Rogo Deploys Multi-Agent AI to Overhaul Wall Street Diligence

Rogo部署多智能体AI重塑华尔街尽调流程 随着AI将耗时数天的交易执行压缩至数分钟，华尔街正面临企业价值反转，各机构必须从依赖人力规模转向可审计的软件系统。 Gabe Stengel详述了Rogo历经数代基础模型演进的发展历程，展示了垂直应用型AI如何重塑一级和二级市场中的常规财务建模、尽职调查与资本配置。他指出，在受监管的金融环境中，透明的可审计性与稳健的软件支撑架构，远比单纯的模型准确率或底层基础智能更为关键。Stengel概述了软件计费从按席位的数据终端订阅向基于交付结果定价的转变，介绍了Rogo内部名为Shrek的AI操作系统，并回顾了在实现规模化扩张前连续遭遇40次A轮拒投的经历。最终，他向华尔街传统机构提出了一个关乎生存的挑战：重构工作流程以实现常规信息分析的自动化，同时将人力带来的阿尔法收益转向专有项目源获取与高信任度关系的构建。 Rogo正在为华尔街构建垂直AI基础设施，推动业务从早期脆弱的提示词链转向能够进行端到端财务分析和跨季度复杂计算的推理模型。 (1:24) 相比单纯的准确率，企业金融更要求绝对的可审计性；机构投资者在信任模型输出之前，需要完整的逐项引用来源、数据链条溯源以及严格的重大非公开信息（MNPI）隔离。 (34:48) 金融机构的软件定价模式正迎来转变，从传统的按席位数据终端收费转向按投资备忘录、审计后的LP报告或卖方CIM收费的按结果定价模式。 (23:43) 在获得Keith Raboy的A轮投资之前，由于资方对模型错误率的疑虑以及硅谷对华尔街软件潜在市场规模（TAM）缺乏认知，Rogo曾连续遭到40次风投拒绝。 (45:00)

Rogo founder Gabe Stengel argues that generative AI will force institutional finance to reinvent its operating models within two to five years. He contends that defensibility in financial software lies not in underlying foundation models, but in vertical-specific execution harnesses, auditable workflows, and private-market transaction plumbing. Targeting Private Market Plumbing Over Public Equities (10:00) While public equities possess abundant structured data and mature automated infrastructure, Stengel focuses Rogo primarily on private equity and investment banking dealmaking. Private markets remain heavily reliant on manual coordination, bespoke negotiation, and unstructured diligence, leaving room to build the core plumbing that links virtual data rooms, due diligence questionnaires, and internal systems of record. By starting with investment banks, software providers also gain access to large seat counts and a direct talent pipeline into buy-side institutions. Stengel asserts that as entry-level analytical tasks are automated, human edge will shift away from standard quantitative calculation toward qualitative judgment and proprietary information gathering. Investors will increasingly rely on boots-on-the-ground legwork, expert networks, and relationship graphs that generate unique inputs base models cannot scrape. The Defensibility of the Execution Harness (16:57) Because frontier AI labs chase platform revenue measured in hundreds of billions or trillions of dollars, Stengel believes they are unlikely to invest engineering resources into the idiosyncratic compliance and operational edge cases of capital markets. A defensible vertical application requires a specialized harness—the tooling, state verification, and context management surrounding the core model—rather than incremental gains in raw model reasoning. This harness must adapt to high-finance habits, such as enabling senior bankers to email redlined decks on mobile devices and receive updated files with traceable revision logs in twenty minutes. Technically, vertical systems have moved away from brittle multi-step pipelines toward unified frontier models equipped with specialized external tools. In regulated financial services, auditability takes precedence over pure accuracy. To satisfy compliance mandates and legal discovery in forums like the Delaware Court of Chancery, platforms must maintain data isolation to prevent the leakage of material non-public information and expose unambiguous reasoning chains for every calculation. Transitioning from Seat Licensing to Outcome Pricing (22:00) Enterprise buyers in high finance currently evaluate vertical AI through traditional seat-based licensing models, mirroring incumbents like Bloomberg and FactSet. Unlike developer-facing software that monetizes direct token consumption without human-in-the-loop sales, selling into institutions requires hands-on sales engineering, security assessments, and high-touch enterprise deployment. Stengel expects this dynamic to undergo a two-phase evolution: first from per-seat fees toward usage tiers, and ultimately to outcome-based pricing. In financial workflows, outcome pricing aligns with concrete deliverables: fees charged per high-conviction investment idea, per verified LP report, or per drafted Confidential Information Memorandum. Long term, Stengel projects that investment workflows will transition from assistive copilots to autonomous agents communicating across firms. This shift could standardize and clear unstandardized private assets, mirroring how retail mortgage underwriting migrated from physical branch visits to automated digital platforms. Navigating Incumbent Vulnerability and Internal Adoption (50:22) Stengel argues that financial institutions face their first true Innovator's Dilemma in decades, as structural fee cushions and fund vintage cycles historically shielded them from technology-driven operational shifts. He poses an enterprise value inversion: if an elite firm currently derives 90% of its value from personnel, it must prepare for a future where 90% of enterprise value is embedded in proprietary software, data, and systems. Institutions are torn between using these efficiencies for head-count margin expansion or entering new markets, such as bulge-bracket banks using single-banker deal teams to profitably cover small-business M&A. Internally, Rogo addresses operational friction and talent scaling through an institutional memory platform called Shrek, which ingests company transcripts to proactively supply context for sales calls. To enforce an AI-native posture, Stengel institutes monthly division audits where tool usage is stack-ranked, humorously placing a dunce cap photo of the lowest-utilizing employee in office spaces to deter organizational complacency. Uncertainty remains regarding the pace at which private markets will embrace automation. While highly standardized transactions like secondaries and private credit can theoretically become autonomous, micro-cap M&A and generational business sales still hinge heavily on interpersonal trust and human relationships that may resist purely digital execution.

Rogo 创始人 Gabe Stengel 认为，人工智能正在以远超以往量化转型的节奏，在未来 2 至 5 年内迫使顶级金融机构重构组织形态；垂直金融 AI 的核心护城河不在于通用大模型本身，而在于连接底层复杂交易、合规系统与工作流的工程“挽具”（Harness）以及细粒度的可审计性。 金融范式转移：从人力转向系统资产 (0:00) Stengel 指出，量化交易巨头 Jane Street 曾耗时约 15 年才建立起基于量化模型的绝对市场地位，但生成式 AI 给现代传统资产管理机构留下的重构窗口期仅有 2 至 5 年。引用 Dario Amodei 提出的“数据中心坐满一国天才”的比喻，高盛、Citadel 或 Millennium 等精英机构的当务之急，在于如何调整旧有的人员层级与业务流程来承接这种海量计算能力。 Stengel 预言，顶级金融机构的企业价值结构将在未来 10 年发生彻底反转：从过去 90% 依赖专业人才，演变为 90% 沉淀在软件、数据与自主系统之中。随着投资备忘录生成与日常分析被全面自动化，纯粹依赖公开信息检索与数据加工的传统分析能力将丧失壁垒；人类投资者未来的核心超额收益，将收缩至非对称的情报搜集能力、排他性的关系网络以及为模型提供独特输入维度的现场洞察。 垂直AI护城河：挽具设计与可审计性 (16:57) 面对通用模型实验室的竞争，Stengel 强调垂直领域厂商必须向基础大模型不愿涉足的纵深发展。OpenAI 与 Anthropic 追逐数千亿至万亿美元规模的市场机会，不会耗费工程精力去适配金融行业碎片化且高风险的合规管道、未公开重大信息（MNPI）管控或虚拟数据室（VDR）。因此，垂直领域的长期壁垒取决于包裹在模型外部的“挽具”（Harness）——即工具集成、执行上下文控制与状态自审环境，这比单纯的基础模型推理微调更能决定企业软件的实用价值。 在架构设计上，Rogo 放弃了早期依赖数十个小模型串联的脆弱管线，转向配备高效工具的大模型自主代理架构。Stengel 提出，在严苛的机构金融环境中，“可审计性比绝对准确性更重要”。由于金融机构面临潜在监管审查甚至特拉华州衡平法院的传票质询，黑盒模型哪怕给出正确答案也会因无法溯源而被弃用；反之，若系统完整呈现推演公式、引用来源与数据血缘，即使偶有假设偏差，也能被机构专业人员快速核验并付诸决策。 战略切入点：攻坚私募市场与成果定价 (10:00) 在市场路径选择上，Rogo 主动避开了数据基础设施完善且高度自动化的公开股票市场，重点切入私募股权（PE）与并购（M&A）等私募交易领域。Stengel 分析称，公开市场具有成熟的交易所和结构化数据流，而私募市场依然充斥着繁重的人工协调、碎片化尽调以及非结构化谈判；解决尽职调查问卷（DDQ）匹配、数据室文档解析与交易系统自动同步，能够撬动巨大的企业级市场空间。 在商业模式演进上，垂直金融软件预计将从当前的按席位收费（Per-seat）过渡到用量计费，并最终进入基于业务成果定价（Outcome-based pricing）阶段——例如直接按高确信度的投资标的推荐、符合审计标准的季度 LP 报告或卖方交易备忘录（CIM）单项收取费用。Rogo 的长远构想是超越人机交互的 Copilot 阶段，构建支持智能体对智能体（Agent-to-Agent）跨机构直接沟通与结算非标资产的底层交易网络。 极限迭代：40次融资被拒与组织大脑 (45:00) 在早期创业阶段，Rogo 在 A 轮融资中遭遇了顶级风投机构连续 40 次拒绝。Stengel 回顾称，这一方面源于硅谷投资人对华尔街垂直软件真实市场规模（TAM）缺乏直观认知，另一方面也因为早期模型能力不足导致测试错误率一度高达 50%。该局面直至投资人 Keith Raboy 决定逆向注资才得以扭转。 为应对底层大模型月度迭代带来的系统淘汰风险，Rogo 效仿 Affirm 每年重构底层分类账的技术思路，持续拆解重构自身的智能体调度框架，并先后收购了 6 家早期金融 AI 团队以吸纳兼备金融常识与工程能力的人才。在内部组织治理上，团队开发了名为“Shrek”的知识中枢，全面转录内部业务通话，结合战略指标实现主动与被动赋能，并辅以内部 AI 工具使用率排名通报，倒逼组织全面适应 AI 原生化运营。 Stengel 指出，当前金融机构已实现显著的个人效率跃升（例如将数天的尽调分析压缩至数十分钟），但真正的瓶颈在于机构顶层战略：究竟是将生产力释放用于削减运营成本，还是利用“一人交易团队”拓展如摩根大通进军中小企业并购等增量市场。最大的未知数依然在于私募市场的非标准化特性与人际信任纽带的持久性——尽管标准化结构资产可逐步实现完全自主的算法匹配，但在高度依赖握手信任与代际交接的长尾微型企业交易中，纯算法结算与自动化定价仍将面临不可预测的阻力与过渡周期。

## Background 该内容由知名商业与投资播客网络 Colossus 旗下的旗舰节目《Invest Like the Best》（第 492 期）于 2026 年 9 月 22 日发布在 RSS 及各音频平台。节目详细介绍了金融 AI 平台 Rogo 的联合创始人兼 CEO Gabe Stengel，重点探讨顶级投资机构如何围绕 AI 全面重塑运作机制，以及 Rogo 如何打造专属金融基础设施以赋能华尔街的交易与分析流程。 ## TL;DR Gabe Stengel 详述了 Rogo 伴随底层大模型演进的历程，强调垂直 AI 胜出的关键并非原始模型智商，而在于解决非标流程的“最后一公里”工程外壳、严密的数据血统以及“可审计性胜过绝对准确率”的信任基石。Rogo 避开前沿实验室的通用赛道，选择深耕流程泥泞的一级市场交易撮合场景，正推动从传统席位制向基于成果定价（Outcome-based）的商业模式演进。此外，Gabe 还分享了公司内部大脑“Shrek”的知识沉淀实践、经历 40 家顶级风投连续拒投的“嚼玻璃”创业历程，并断言垂直 AI 赢家必须保持极致进攻性，迅速成为吸纳顶尖人才与资本的超级黑洞。 ## Atmosphere & style 整期对话节奏凌厉、信息密度极高，兼具华尔街交易员的务实老练与硅谷开拓者的偏执野心。Gabe 毫不掩饰创业过程中的脆弱与至暗时刻，从被顶级风投接连拒绝的刺痛，到对复利势能一旦熄火的神经质恐惧，展现出罕见的真实与坦诚。整场对谈让听众仿佛置身于激战正酣的应用层 AI 决胜现场，既能直观体会到前沿技术撞击传统金融体系的巨大冲击力，也能深刻感受到创始团队在刀尖上狂奔的紧迫感与狼性斗志。 ## Bulletin - `01:24` Rogo 的产品能力随前沿基础模型经历质变：o1 Pro 解决了金融检索与指标计算的稳定性，而 Opus 4.5 时代则已能全面胜任初级投行分析师的基础工作。 - `38:15` 在严肃金融与资本市场场景中，可审计性（Auditability）的重要性压倒了绝对准确率（Accuracy），透明可复查的推导链路是让金融机构敢于执行真实交易的唯一信任基石。 - `24:25` 垂直 AI 企业必将经历从席位制定价、用量计费到基于成果定价（Outcome-based pricing）的演进，直接按生成的投研洞见、投后报告或推介备忘录（CIM）收取费用。 - `35:49` Rogo 团队内部通过录音全员会议构建了名为“Shrek”的知识大脑以加速新人赋能，并通过激进收购 6 家 AI 初创公司补齐了复合型产品与工程人才拼图。 ## Key takeaways - `01:24` 未来 2 到 5 年内，全球顶尖投资机构都将被迫根据 AI 重构业务流程，AI 全生命周期的缝合难度堪比当年量化巨头耗费 15 年构建的体系壁垒。 - `04:15` 垂直金融 AI 的核心壁垒不在于调用通用大模型，而在于打通深水区的“最后一公里”工作流，如 MD 的 iPad 批注转 Deck 与严密的审计追溯。 - `10:00` Rogo 选择从由人工驱动、缺乏数字化基础设施的一级市场交易撮合者（Dealmakers）切入，而非高度自动化的二级公开市场。 - `10:55` 当 AI 逐步商品化基础建模与 IC Memo 撰写等初级分析技能时，人类投资人的核心护城河将退守至敏锐的裁量权（Judgment）与专有关系网络图谱。 - `16:57` 与前沿基础模型实验室（如 OpenAI、Anthropic）必须走正交错位竞争，深耕满足 MNPI 监管合规、数据室管理等泥泞定制化的垂直痛点。 - `20:00` 胜出的垂直 AI 需具备三大要素：行业极高的数据与老旧系统复杂度、团队深谙业务且充满好奇心，以及敢于推倒重构核心产品的工程魄力。 - `22:50` 底层大模型目前最大的演进瓶颈在于长上下文的“记忆压缩”，这一能力缺失在数百人规模的企业协作网络中会被成倍放大。 - `28:10` 长远来看，资本市场将演进出类似线上房贷审批的自动化即时风险定价机制，大幅降低资产定价的时间量级并释放长尾流动性。 - `30:00` Rogo 的终极构想是超越 Bloomberg Messenger 式的人际沟通，构建一套支持跨机构 AI Agent 自主协商与资产流转的数字化交易所网络。 - `31:55` 优先攻关大型投行不仅能获得极高客单席位密度，还能将品牌心智深植于向买方流动的年轻分析师网络中，形成强大的人才分发杠杆。 - `33:45` 基层员工正借助 AI 演变为“一人交易团队”，使原本佣金微薄的中小企业长尾并购业务具备了极高的商业回报可行性。 - `43:00` 红杉合伙人 Pat Grady 提醒创业者必须采取极致进攻策略：在全行业 18 个月内必做 AI 采购决策的窗口期，不惜推高失败风险也要主动拉宽千亿市值的概率两端。 - `45:00` 早期连续遭遇 40 家顶级 VC 拒绝源于硅谷对金融 TAM 的认知盲区、模型迭代初期的缺陷，以及资方对创始人持续“嚼玻璃”坚韧度的怀疑。 - `53:09` 金融掌门人必须面对企业价值载体的颠覆：未来机构 90% 的价值将不再依附于合伙人个人能力，而是沉淀在软件系统、专有数据与数智大脑中。 - `59:37` 垂直 AI 创业是一场严酷的职业联赛淘汰赛，企业必须在短时间内蜕变为虹吸顶尖人才、海量资本与强大品牌的超级黑洞，才能避免被底层模型实验室吞噬碾压。 ## Detailed walkthrough ### `01:24` 投资超级智能与 Rogo 的模型演进时代 对冲基金与投资机构的核心职能正在被技术迅速重构。面对“AI 是否正在吞噬优秀分析师乃至投资组合经理（PM）核心工作”的疑问，Gabe Stengel 认为，预测未来两年的演进比推演一二十年更为清晰：未来两到五年内，全球最顶级的投资者都将被迫思考如何利用 AI 彻底重塑公司运作模式与自我定位。这种范式转变犹如当年量化交易与做市业务的崛起——Jane Street 耗费了整整 15 年才建立起绝对的统治壁垒。正如 Anthropic 创始人 Dario Amodei 的经典论断所言，“未来每个机构的数据中心里都会装着一个国家规模的天才群体”。如果高盛（Goldman Sachs）、千禧年（Millennium）或城堡投资（Citadel）突然迎来一个由顶尖天才组成的国家协助办公，即便强如这些华尔街巨头，也需要耗费数年时间去调整业务流程、摸索协作机制并将其融入现有体系。因此，如何将 AI 真正缝合进投资的全生命周期，是所有顶级投资者在未来五年面临的最大挑战。 回顾 Rogo 的进化历程，其产品形态经历了痛苦的技术跃迁，与底层基础大模型的迭代节点深度绑定。在真正创立 Rogo 之前，Gabe 曾经历过两次失败的尝试：第一次是在高中时期，一位投行家父亲希望开发一款能够追踪两家上市公司并购时股权交换比率的软件，当时使用的古老 AI 技术效果极差；第二次是在大学期间，他在 GPT-3 问世前发表了一篇关于计量经济学与金融计量 AI 助手的学术论文，并试图将其商业化，最终同样以完全失败告终。直到 GPT-3 诞生、ChatGPT 尚未引爆的前夕，Rogo 正式创立。早期的产品虽然演示看起来很惊艳，但实际根本无法稳定运行。在应用层 AI 领域，先行者往往背负着巨大的劣势——你清晰地预见了未来的终局并提前构建产品，但底层模型能力却尚未就绪，早期用户试用后只会给出“糟糕透顶的垃圾”这种负面评价。 Rogo 真正走向实用化，本质上是由两次关键的模型跃迁驱动的： - **o1 Pro 时代**：模型首次展现出足够高的数据可靠性，至少成长为一个合格的金融检索工具。用户可以输入指令要求其计算某家公司过去 12 个季度的特定财务指标，模型能够极其稳定地输出结果，不再频繁出错导致用户抓狂到宁愿自己去算。 - **Opus 4.5 时代**：在跨过模型能力的临界点后，只要给予正确的上下文和精确指令，模型已经能够胜任初级投资专员或初级投行分析师所负责的绝大多数基础工作。一旦底层模型追赶上产品设计的终局构想，整个产品的体验便瞬间产生质变，客户开始反馈产品帮他们每个月省下了数百小时，让决策更加精准敏锐。 ```mermaid flowchart LR accTitle: Rogo模型演进与能力跃迁历程 accDescr: 展示Rogo从早期失败尝试到o1 Pro以及Opus 4.5驱动下的能力进化路径 subgraph EarlyStage["早期探索阶段"] A["高中与大学尝试<br/>传统AI / 预训练前技术<br/>(完全无法落地)"] --> B["GPT-3 早期时代<br/>Demo表现惊艳<br/>实际业务中完全不可用"] end subgraph ModelLeap["大模型能力突破"] B --> C["o1 Pro 时代<br/>金融搜索与指标计算稳定<br/>解决基础财务检索痛点"] C --> D["Opus 4.5 时代<br/>上下文完备时胜任初级投行工作<br/>赋能复杂投资全生命周期"] end ``` --- ### `04:15` 最后一公里、合规与投资银行家的工作流细节 应用层 AI 产品的核心壁垒不仅在于调用模型，更在于对行业“最后一公里”工作细节的深刻洞察与极致打磨。金融机构对合规审查、监管流程以及交付物格式有着严苛的要求，但打动顶级金融从业者的往往是那些极其细微的人性化工作流设计。以投资银行董事总经理（MD）的典型日常为例：许多资深 MD 已经长达十年没有真正坐在电脑前完整操作复杂软件，但他们非常习惯随身携带一台 iPad，在演示文稿（Pitch Deck）上直接用手写笔圈改标注。传统的流程是 MD 将批注扫描或拍照后通过邮件发给初级分析师，分析师通常需要熬夜加班一到两天才能修改并交付新版本。 Rogo 针对这一现实场景重构了人机交互路径：MD 依然保留原有的工作习惯，只需将标注好修改意见的 Deck 像往常一样发封邮件，但收件人变成了 Rogo 的 AI 分析师。系统在 20 分钟内便能完成全套修改并回传，同时自动抄送项目组内的初级分析师。系统不仅快速生成结果，还会向初级分析师完整展示所有细微改动的可审计追溯路径，以便人工随时介入复核与确认。这种产品设计完全尊重了行业老兵的直觉与习惯，将原本耗时数天的机械劳动压缩至数十分钟，既消除了繁琐的技术使用门槛，又通过严密的审计链路保证了金融输出的严肃性。 在技术前沿探索上，金融领域展现出了与其他行业截然不同的代币（Token）经济学特征。普通行业很难承受为了提炼单点信息而消耗海量 Token，但顶级对冲基金的投资组合经理（PM）却完全愿意为“一个真正优质的投资想法”掏出 5 万美元。因此，Rogo 正在探索一种类似“多智能体网络（Multi-agent System）”的前沿形态：为每位 PM 配置上万个后台智能体，让它们全天候阅读财报、推演研报、相互质询与激烈辩论，在经过连续 24 小时的多轮碰撞后，最终仅向决策者提炼输出一条极具深度的投资洞察。 --- ### `06:14` 谁在真正使用 Rogo：交易撮合者与冰山之下的系统集成 虽然公开市场股票交易（Public Equities）表面上看是 AI 最理想的试验场——因为所有公开数据唾手可得，看似只需凭借算力进行“暴力破解”与智能比拼即可——但 Rogo 目前最为核心的理想客户画像（ICP）与主力用户群体，实际上是负责实际交易撮合的“交易者（Dealmakers）”，即那些从事公司买卖、收并购并协同复杂交易流程的投行家与私募股权（PE）投资机构。一笔真实的交易需要跨越漫长且繁琐的阶段：搭建数据室（Data Room）、解析第三方数据室、协调中介机构电话会议、解答尽职调查问卷（DDQ），直到最终完成交割。 在实际使用场景中，买卖双方都在高频运转 Rogo 的能力： - **卖方团队**：利用 AI 准备全套交割材料，包括自动生成并校验公司的财务模型、撰写分析客户画像与留存率的汇报 PPT、批量填写尽职调查问卷（例如详细拆解客户集中度风险等问题）。 - **买方团队**：利用智能体集群反向穿透卖方数据室中的海量文档与底层数据，并自动将其与本基金的投资哲学和风控框架进行比对，快速推导该标的的客户集中度是否处于该期基金设定的风险容忍阈值之内。 ```mermaid flowchart TD accTitle: Rogo在交易全流程中的双向赋能与底层系统交互 accDescr: 展示卖方生成材料、买方解析数据，以及Rogo作为水下冰山对接记录系统的全景 subgraph TransactionFront["交易前台交互"] Seller["卖方团队 (Sell Side)<br/>生成财务模型 / 客户PPT / DDQ问答"] DataRoom[("虚拟数据室 (Data Room)")] Buyer["买方团队 (Buy Side)<br/>智能体穿透数据 / 比对投资哲学与风控标准"] Seller -->|沉淀与上传材料| DataRoom DataRoom -->|解析与尽调查验| Buyer end subgraph RogoIceberg["Rogo 水下冰山 (记录系统集成)"] CRM["核心 CRM 系统 (更新交易进展)"] Portfolio["投后组合监控系统 (动态调整指标)"] LPReporting["LP 报告分发系统 (自动同步数据)"] end Buyer -.->|自动写回与双向同步| RogoIceberg Seller -.->|自动写回与双向同步| RogoIceberg ``` 除了常见的聊天机器人界面、邮件交互与主动风险预警外，Rogo 超过一半的业务价值实际上隐藏在“水面之下的冰山”中。在极其看重规范的大型机构内，交易人员在推进项目的每一步时，都必须同步更新后端的各项核心“记录系统（Systems of Record）”。Rogo 深入集成了机构的底层系统，能够在交易员推进业务的同时，自动更新内部 CRM 系统中的项目状态、同步调整投后组合监控指标，甚至将结构化信息沉淀并直接分发至后续面向出资人（LP）的报告系统中，从而彻底消除了传统交易员在不同业务系统间反复手动搬运数据的沉重负担。 - **多端触达方式**：涵盖电子邮件自动响应、iPad 批注流转、原生 Web 对话界面以及主动式数据异动告警。 - **后台系统自动写回**：深度串联企业 CRM、资产组合风控数据库以及 LP 投资者关系管理平台。 - **业务场景覆盖度**：横跨投行 M&A 卖方材料筹备、PE 买方尽职调查拆解，以及横跨多组织的多方交易协调全流程。 ### `10:00` 一级市场与二级市场：为何从“交易撮合者”切入 许多人的直觉通常认为二级公开市场（Public Markets）拥有海量且规范的公开数据，本应是应用人工智能的最佳试验场。然而，Gabe Stengel 坦言，要打造真正顶级的公开市场投资工具，必须具备极其深厚的垂直领域认知，而他此前并未在一级以外的二级公开市场实战过。尽管他坚信 Rogo 所搭建的底层系统和基础设施未来能为公开市场创造巨大价值，但当下需要找到一位“顶级主厨”，将这些底层技术组合调配出适合二级市场投资者的终端产品与最后一公里交付体验。 相较于二级市场已有成熟且自动化的交易所和数据管道，一级私募市场（Private Markets）几乎完全由人工驱动。无论是不同机构间的协同、数据标准化、尽职调查还是最终的交易撮合，所有环节都极度依赖人力，存在大量尚未搭建的基础水暖管道（Plumbing）。尽管董事会屡次敦促 Gabe 扩展其理想客户画像（ICP），但他始终坚持深耕“交易撮合者”（Dealmakers）这一群体。在他看来，二级市场的成熟基础设施早已成型，而一级市场充满未被数字化的泥泞流程，正是垂直 AI 构建底层交易基础设施的绝佳土壤。 - **基础设施差异**：二级公开市场已高度自动化并拥有成熟交易所；一级私募市场仍高度依赖人工协调、信息标准化与离线交易。 - **垂直 AI 策略**：在“水暖管网尚未建成”的垂直领域，先搭建基础设施并叠加专有数据，具备远高于纯工具软件的护城河。 - **TAM 深度**：仅服务于精明的银行家与交易撮合者，其潜在市场规模与客单价值就足以支撑起极其庞大的商业体。 --- ### `10:55` 投资能力的重塑：从初级分析到“第37手”与专有数据 当 AI 能够端到端完成从数据收集、尽职调查、撰写投委会备忘录（IC Memo）到处理异议等整套初级分析师工作时，投资专业人士究竟该依靠什么能力立足？Gabe 谦逊地表示自己仅在金融业工作过两年，比起技术极客，他更像一名虚心向顶级投资人请教的学生。他认为，世界顶级投资人的核心优势在于能够在海量噪音中敏锐判断出真正重要的议题，并在不确定性中展现出卓越的个人裁量权（Judgment）。 对于 AI 是否能彻底替代人类的投资判断，Gabe 提出了一个耐人寻味的类比：就像 AlphaGo 在对战李世石（Lee Sedol）时下出了震动围棋界的“第37手”（Move 37），走出了人类从未见过的神之一手。如果在二级市场上 AI 也开始频频展现出这种超越人类认知的决策洞察，整个投资能力的衡量标准将被彻底颠覆。在那个未来，人类投资人最核心的稀缺技能不再是坐着计算模型，而是深入一线、拜访专家，构建起属于自己的关系网络图谱（Relationship Graph），从而为模型输入全世界独一无二的专有数据与事实。 ```mermaid flowchart TD accTitle: 投资专业技能价值的演进路径 accDescr: 展示从传统的分析师硬技能向未来的专有数据输入与独特关系网络转化的逻辑。 subgraph Legacy["传统技能（逐步被 AI 商品化）"] A["财务建模与数据清洗"] B["撰写 IC 备忘录"] C["常规尽职调查与异议整理"] end subgraph Frontier["核心稀缺技能（价值持续放大）"] D["敏锐判断力 (Judgment)"] E["一线实地调研与专家访谈"] F["独家关系网络图谱 (Relationship Graph)"] end Legacy -->|"AI 接管端到端流程"| G["垂直 AI 智能体 (Rogo)"] Frontier -->|"喂入独家外部变量"| G G -->|"辅助或生成"| H["'第37手'级别的超额决策"] ``` --- ### `15:00` Rogo 数据与模型栈的进化：告别“鲁布·戈德堡机械” 回顾 Rogo 早期系统的构建，Gabe 形容那简直就像一台脆弱复杂的“鲁布·戈德堡机械”（Rube Goldberg Contraption）。当用户抛出一个金融问题时，系统内部需要链式调用多达 60 次模型：先识别涉及哪家公司，再推断股票代码，接着用代码调用彭博（Bloomberg）或 FactSet 的 API 获取数据，最后再换另一个模型进行汇总拼接。然而随着基础模型变得越来越聪明，系统的设计哲学从死板的规则设定，转向为智能体提供最简洁、最高质量的工具箱。 现代 Rogo 架构如同在培养一名刚入职但智商极高的顶尖投资人，重点在于赋予其自身智商之外的外部工具链： 1. **数据抓取与工具调用**：构建针对金融多模态数据的查询与信息搜集工具。 2. **自我审计机制**：确保计算流程具备完整的因果链条，不仅能做分析，还能回溯校验。 3. **输出系统集成**：将结果精准格式化并推送回投行与基金正在运行的核心业务软件中。 4. **合规隔离与溯源**：提前布局监管合规，确保未来即使面对特拉华州衡平法院（Delaware Court of Chancery）的证据开示，AI 输出的每一项投资判断与研报都能清晰回溯数据血统（Data Lineage），且绝不发生机构间的信息混杂。 5. **动态模型路由与评测**：内部搭建专有评测集（Evals），在表现性能、Token 成本以及响应延迟之间做全局权衡，将不同复杂度的任务分发至最适宜的模型。 --- ### `16:57` 与巨头错位竞争：为什么前沿实验室不会涉足金融脏活 面对“OpenAI 和 Anthropic 也盯上了金融大蛋糕，Rogo 凭什么竞争”的经典质疑，Gabe 的战略非常明确：必须与前沿模型实验室走“完全正交”（Perpendicular）的技术与产品路线。有些初创公司做个聊天机器人（Chatbot）是为了迅速推向市场，但这只是浮在水面上的冰山一角。金融服务与资本市场中蕴藏着足以诞生年营收 50 亿至 100 亿美元垂直巨头的广阔空间，其本质是大量极其混乱、定制化且壁垒森严的专有痛点。 > "对于 Anthropic 来说，做这些事情就像走在通往万亿美元营收的大道上，停在路边去弯腰捡一分钱硬币；但对我们而言，这正是构筑数百亿美元垂直帝国的核心支撑。" 金融行业的复杂性绝非单纯提升智能水平就能解决。以处理重大非公开信息（MNPI）为例，当金融机构处理影响市场走势的并购交易时，合规、审计及内部监控要求苛刻到“像素级”（Pixel-perfect）。这就要求系统不仅是一个分析引擎，还必须是一个绝对合规、安全的虚拟数据室（Virtual Data Room），并与智能体及具体工作流无缝缝合。Anthropic 和 OpenAI 绝不可能沉下心去构建这种沉重繁琐的数据室业务。如果一家公司的终局目标是成为高阶金融与资本市场所有交易的“数字交易所”，就必须同时掌控底层智能、交易场所、沟通渠道、执行工作流以及所有被摄入的核心数据。 - **战略定位**：前沿实验室追求从 1000 亿跨越到万亿美元的通用智能（AGI）底座，垂直企业则深耕金融各细分领域的专有标准与监管管道。 - **不可替代的泥泞细节**：MNPI 审计追溯、受监管的数据室（Data Room）、机构级安全风控流程。 - **基础设施与外壳（Harness）的重要性**：如同 Claude Code 与协作界面展现出的体验跃升，决定垂直领域 AI 成败的往往不是底层基础模型本身的细微差异，而是包裹在模型外部的运行外壳（Harness）与工程基础设施。 ### `20:00` 驾驭模型与垂直领域 AI 的胜出法则 单纯拥有原始模型能力与庞大参数量并不足以构成真正的应用壁垒，关键在于如何对模型进行工程化驾驭（Harness）。人类之所以具备极强的主动性（Agency）与解决复杂问题的能力，不仅源于高智商或记忆力，更在于大脑中协同运作的各种“微服务”机制——如何归档知识、如何精准检索、如何通过情绪与本能触发反应。顶尖投资人之所以拥有敏锐的判断力与“直觉天线”（Spidey sense），正是因为市场异动能够瞬时唤醒过往经验微服务，从而做出创造性的投资决策。垂直 AI 应用的核心任务，正是为行业构建这套复杂的微服务架构。 如果从投资人的视角审视垂直领域 AI 赛道，能够最终脱颖而出的企业通常具备三大硬性特质： 1. **行业具备极高的数据复杂度与系统深度**：该行业必须充斥着独特的复杂数据类型、老旧的记录系统（Systems of Record）以及繁琐的私有化部署要求。如果一个行业无需复杂集成、任何人套用开箱即用的 ChatGPT 或 Copilot 就能解决问题，初创团队就根本没有足够的时间窗口在垂直赛道建立防御纵深。 2. **团队兼具领域认知与狂热好奇心**：团队必须对所解决的场景抱有极高热情。加布·斯坦格尔（Gabe Stengel）虽然起初并非资深金融家，但自幼成长于纽约，身边充斥着投行与资本市场的话题。Rogo 组建了一支由 100 多名拥有顶级投行及头部投资机构背景的专业团队，专门负责捕捉前沿模型的每一次迭代，并将其严丝合缝地驾驭到金融机构的实际业务中。 3. **敢于彻底推倒核心产品的迭代魄力**：团队必须时刻准备“亲手摧毁现有产品并从零重构”。如果一家公司的交付界面被固化在某个特定终端（Terminal）或交互框架中无法自拔，它就注定无法跟上基础模型每六个月一次的代际跃迁。这类似于 Affirm 创始人麦克斯·列夫琴（Max Levchin）的工程哲学——每年重构一次核心底层账本系统：一方面以极具挑战性的工程难题吸引并留住顶尖人才，另一方面彻底杜绝系统僵化。Rogo 对其底层智能体架构（Agentic system）也贯彻了同样的理念，在模型飞速演进的当下主动推倒重来，绝不在局部最优解上固步自封。 --- ### `22:50` 模型的最大瓶颈：记忆压缩与多智能体协同 当前底层大模型最欠缺、一旦突破将彻底颠覆产品形态的核心能力是“记忆压缩”（Compaction）。当一名用户与单一智能体进行了上百次对话之后，智能体如何精准提炼出核心偏好与关键记忆，将其压缩进有限的 Token 窗口，同时依然保持对用户意图与背景的连贯认知？若能做到这一点，AI 才会真正让人感觉像是一个不断增进了解、具备长期上下文记忆的真人助手。 这一工程难题在协同网络中还会呈指数级放大： - **单对单交互局限**：目前的 ChatGPT、Copilot 或 Gemini 绝大多数仍停留在单人对单机器人的点对点交互阶段。 - **跨组织网络协同**：一旦智能体需要常驻在拥有上百人的企业级通信渠道（如 Slack 频道）中，与众多团队成员协作，记忆压缩的难度立刻爆发——它必须同时与上百人对话，在不同信息源之间穿针引线，并跨越长周期交互维持一个持久、连贯的全局“组织心智模型”（Mental model）。目前的基础模型在这一能力上仍存在明显短板。 --- ### `24:25` 商业模式之辨：席位制、Token 消耗与结果付费 当下的 AI 软件公司正分化为完全不同的商业范式。一类是“Token 代理商/消耗型模式”，典型代表如面向工程师的 Cursor、Factory 或 Claude Code，这类工具面向习惯于用量计费的开发者群体，无需庞大的人力销售即可实现商业收入的指数级飞跃；而 Rogo 则属于重度企业级软件（Enterprise sales）。 由于华尔街买方的预算习惯是将 Rogo 与彭博（Bloomberg）、FactSet、S&P Capital IQ 及 PitchBook 等传统金融终端归为同类，因此 Rogo 目前仍然采用按席位付费（Seat-based pricing）的商业模式。这就要求公司建立极其传统而重度的 GTM（Go-to-market）铁军：每一个大单落地，都需要客户经理（AE）与解决方案架构师（Solutions Architect/Sales Engineer）深入现场，与金融机构内部的各方利益相关者打交道、解释技术原理并协助系统集成。 ```mermaid flowchart LR accTitle: AI 软件定价模式的演进路径 accDescr: 展示从席位制到用量计费再到结果付费的商业模式演进。 subgraph S1["传统阶段"] A["席位制定价<br/>(Seat-based)"] end subgraph S2["过渡阶段"] B["用量/Token消耗<br/>(Usage-based)"] end subgraph S3["终局阶段"] C["结果/价值定价<br/>(Outcome-based)"] end A -->|逐步打破人头限制| B B -->|跳过算力分歧| C ``` 加布认为，所有垂直 AI 企业最终都必须跨越两场定价革命：先从席位制走向用量计费，再最终迈向“基于成果的定价”（Outcome-based pricing）。Rogo 的长期策略甚至希望直接跳过充满争议的 Token 用量计费模式： - 客户如果消耗了 10 万美元的 Token，往往会对实际获得的价值产生分歧与疑虑。 - 但如果直接按产出成果定价，价值衡量就一目了然：为买方生成一个被采纳的高质量投资洞见（Investment idea）收取费用；为基金向出资人（LP）生成一份精准合规的季度投后报告收取费用；或者为投行撰写一份完整的企业推介备忘录（CIM, Confidential Information Memorandum）收取固定费用。 面对当前按席位收费下“客户使用频率越高、自身模型调用成本越大”的短期错配，加布表现得相当坦然：“Rogo 目前仅完成了产品路线图的 1%，资本市场 99% 的创新空间还在前方。”现阶段最核心的考量是成为客户在 AI 战略上最值得信赖的长期伙伴，而非计较单个席位的算力边际成本。 --- ### `28:10` 资本市场的终局：从人工瓶颈到自动化风险定价 纵观现代资本市场，绝大多数投资与并购流程、投行交易机制，依然严重受制于人类处理信息的速率极限（Human rate limited），并依赖昂贵的人类中介网络运转。这一业态正在复刻 15 至 20 年前住房抵押贷款领域的演进历程：彼时人们普遍认为申请房贷必须亲自走进线下网点面见信贷员，而如今通过 Rocket Mortgage 等数字化平台，40% 至 50% 的房贷审批在线上自动化完成。 未来的资本市场生态将迎来全方位的重构： - **融资门槛平民化**：未来企业创始人或业务主管发起数百万美元的股权或债权融资，将如同散户在 Robinhood 上点击买入股票一样顺畅。 - **交易决策极速化**：像 KKR 这样的私募股权巨头评估是否将旗下某家被投企业出售给另一家赞助商，过去需要长达五个月的繁琐尽调，未来通过智能体系统将在五分钟内完成意向锁定。 - **全资产的“即时风险定价”**：整个金融世界将演化出类似房产即时交易平台（如 Opendoor）的自动化风险定价机制，大幅降低资产定价的时间量级，释放难以估量的流动性与交易透明度。 为了实现这一未来蓝图，Rogo 正在复制彭博当年的经典三步走扩张飞轮： ```mermaid flowchart TD accTitle: Rogo 的资本市场演进飞轮 accDescr: 展示从数据切入到分析工作流，最终成为交易与通信平台的彭博式扩张路径。 subgraph Phase1["第一阶段：破局切入"] P1["专有数据与基础接入<br/>(Data Wedge)"] end subgraph Phase2["第二阶段：深度绑定"] P2["分析引擎与智能体工作流<br/>(Analytics & Workflows)"] end subgraph Phase3["第三阶段：资本终局"] P3["交易所与交易通信平台<br/>(Exchange & Transaction)"] end P1 --> P2 P2 --> P3 ``` - **第一阶段（破局切入）**：以专业金融数据和特定模型能力切入投行与买方的工作流； - **第二阶段（深度绑定）**：在底层数据之上构建从业者日常依赖的完整分析、建模与交付工作流； - **第三阶段（终局平台）**：顺理成章演变为机构间沟通、撮合交易与资产流转的基础设施交换网络。 ### `30:00` 从 Copilot 到 Autopilot：构建跨机构的 Agent 交易网络 在谈及非标资产与不透明市场的演进路径时，Gabe Stengel 将 Rogo 的愿景与金融行业历史上的基础设施进行了对照。彭博即时通讯工具（Bloomberg Messenger）当年本质上是为人类交易员搭建了一个撮合与沟通的渠道，使得过去模糊不透明的各类资产交易得以在线上流转。而 Rogo 的切入策略是：先用轻量级的 AI 能力敲开客户的大门，随后逐步接管端到端的全流程工作流，完成从“Copilot 聊天机器人”向“全自动 Autopilot 工具”的跨越。这种跨越不仅意味着系统能在投资委员会备忘录（IC Memo）或尽调问卷（DDQ）上实现 100% 的精准度，更在于它能够充当交易对手方之间的全新通信信道。 与 Bloomberg Messenger 截然不同的是，Rogo 不需要搭建一个让人类彼此对话交易的平台，而是要构建一套让穿梭于各大金融机构和投资机构之间的“AI Agent”进行互联、协商并完成交易的基础设施。Gabe 描绘了一个未来的场景：大型私募股权基金（PE）如果敢于放手让一个 Agent 代表自己去谈判交易，系统就必须能够自动与交易涉及的所有第三方顾问进行无缝对接——包括负责收益质量分析（QoE）的会计师、律所的法务顾问等，并能组织协调一场由多家财务赞助商（Sponsors）提交竞标方案的完整拍卖流程。要实现这一构想，底层需要搭建海量的复杂软件系统，这在本质上相当于为那些从未标准化的金融资产重构一座现代化的“数字化交易所”。 ```mermaid flowchart TD accTitle: 从人际通信到自主 Agent 交易网络的演进 accDescr: 展示金融工作流从依赖人工的 Bloomberg Messenger 到 AI Copilot，再到最终全自动 Agent 间跨机构撮合交易的演进链路。 subgraph Phase1["第一阶段：人际沟通"] A["Bloomberg Messenger"] --> B["人类交易员手工撮合"] end subgraph Phase2["第二阶段：单点辅助"] C["AI 助手 / Copilot"] --> D["自动化底稿与 IC Memo 生成"] end subgraph Phase3["终极形态：跨机构 Agent 网络"] E["PE 买方 Agent"] <--> F["Rogo 跨机构通信协议与交换基础设施"] G["第三方顾问 (QoE / 法务)"] <--> F H["卖方 / 竞标赞助商 Agent"] <--> F end Phase1 -.-> Phase2 -.-> Phase3 ``` --- ### `31:55` 客户布局与破局点：为什么先攻大型投资银行 当被问及当前客户结构在巨型商业/投资银行与买方投资机构之间的分布时，Gabe 透露目前客户仍以大型投资银行为主。这一客群选择直接源于他自己的职业经历——在创立 Rogo 之前，他曾在投资银行从事买方并购（Buy-side M&A）业务，对投行内部的运作痛点有着切身体会。但更深层的商业考量在于两大结构性优势： 首先，顶级投行是整个金融行业最核心的人才分发管道。在整个华尔街，日后走向顶级对冲基金或 PE 机构的大批明星投资人，职业生涯前两年往往都是在高盛（Goldman Sachs）的 TMT 组或类似的一线投行大组中担任分析师。一旦 Rogo 在这一阶段捕获了他们的使用习惯，这些年轻分析师在跳槽到买方机构后，就会自然而然地成为 Rogo 最忠诚的传播节点与倡导者。 其次是极高的单客坐席密度。如果攻克一家像美银（Bank of America）这样体量的超级投行，软件可以迅速分发至成千上万名专业人员手中；相比之下，即便拿下十家业内最顶尖的单基金经理（Single-PM）公开市场股票基金，每家机构可能也仅有十位核心投资人员，覆盖面和扩张速度完全无法相提并论。 - **分发优势**：投行分析师是买方投资机构的摇篮，品牌心智可随人才流动跨机构渗透。 - **席位规模**：大行单次签约即可覆盖海量席位，获客杠杆远高于小型精品买方。 --- ### `33:45` 个人效率爆发与机构战略困境：“一人交易团队”重塑金融供给 谈到当前企业应用 AI 的普及深度，Gabe 坦言整个行业目前正处于一个微妙的阶段：绝大多数机构都已经见证了极其惊人的“个体生产力提升”，但管理层当下最痛苦的挑战在于，如何将这种个体的效率红利转化为账面上可量化、可持续的“机构级生产力”。 在微观层面，AI 带来的震撼已无可置疑。凡是部署了 Rogo 的银行，基层员工的反馈近乎狂热，纷纷表示自己现在的效率是过去的“100 倍”。甚至有董事总经理（MD）直接对 Gabe 感慨：“我刚给客户发了一份 5 页的高质量分析材料。如果在过去，我必须和分析师来来回回沟通修改整整三天；而现在，我自己只花了 10 分钟就全部做完了。”更具讽刺意味又极为真实的是，这些在华尔街身居高位的 MD 们，很多已经整整 20 年没碰过 Excel，25 年没亲手做过基础底层分析，而生成式工具却让他们跨过了所有操作门槛。 然而在宏观层面，企业面临着深刻的战略抉择：这些省下来的时间究竟该流向何处？是能帮你赢下更多 Deal，还是促成了更多实际交易？是借此缩减人力成本，还是去开拓过去根本不合算的长尾业务？Gabe 以摩根大通（JPMorgan Chase）近期的战略举措为例：小摩公开宣布将发力进军中小企业（SMB）并购市场。在过去，这部分长尾市场的交易佣金过于微薄，银行根本无法负担庞大的项目团队成本；但如果借助 AI 工具让单个银行家变成一支“一人交易团队”（Deal team of one），原本经济上不可行的业务板块瞬间就有了极高的商业回报。技术在无限制地释放产能，接下来的瓶颈已经不再是 AI 的算力，而是机构管理者自身的想象力与战略重构能力——正如 Gabe 所比喻的：“如果明天突然有 100 位顶级投资人加入你的麾下，你光是想清楚如何给他们分工、分配多少资金、主攻哪些领域，就得花上好一阵子。” --- ### `38:15` 可审计性胜过绝对准确率：AI 走向自主执行的信任基石 两年前，企业客户在评估 AI 时几乎都在死磕“准确率”以及“模型幻觉（Hallucinations）”，但如今在专业对话中，“幻觉”这个词几乎很少被提及。面对金融领域对小数点后精确度的苛刻要求，Gabe 提出了一个反直觉但至关重要的洞察：“在严肃金融场景中，**可审计性（Auditability）甚至比绝对准确率（Accuracy）更为重要**。” 这两者虽然常被混为一谈，但在实际业务交付中逻辑截然不同： 1. **黑盒的高准确率毫无价值**：如果系统直接抛出一个答案，却无法展示它是如何计算出来的、引用了哪些数据源，那么哪怕该答案实际上 100% 正确，专业金融人员也绝不敢在实际业务中使用它，因为缺乏核查机制就无法建立信任。 2. **完全可追溯的系统具备极高行动力**：即使系统给出的输出偶有偏差，但只要它能清晰透明地列出所有底层假设、推导逻辑和数据抓取来源，分析师就能在几秒钟内快速定位问题并予以修正。此时，它依然能为团队节省大量工时，具有极高的实操价值。 Gabe 强调，随着软件从单纯用于检索信息的 Copilot，进一步迈向被赋予决策权、能直接代为执行交易和发送指令的 Autopilot，这种透明可见的核查链路与可审计机制，将是用户敢于把真实资金与商业决策交给 AI 的唯一前提。 ### `35:00` 可审计性、监管要求与企业级赋能的本质 当 AI 代理在投资或金融业务中替人类做决策甚至发送邮件时，完整的“可追溯性与可解释性”（Auditability）是不可逾越的底线。人类不仅需要对系统抱有充分信心，还必须在出现偏差时能精准复盘——就像某位投资经理做出了糟糕的投资决策时，团队必须调查清楚到底是底层数据输入有误、有人说了谎，还是逻辑推演崩塌一样，AI 智能体同样需要接受这种调试。尤其是在受到监管严格审查的资本市场，一旦无法清晰解释做出某项决策的数据依据和推理链路，任何方案都无法被合规机构接受。 在商业化落地层面，尽管 Rogo 面临着与传统企业级 SaaS 相似的销售与交付流程，但其增长速度却需要比上一代最顶尖的企业级软件快上数倍。不同于 Anthropic 等模型厂商可以通过开放 API 实现使用量指数级激增，应用层软件严重受到人类交付能力的物理限制。Gabe 指出，解决这一瓶颈的核心矛盾在于：如何以最快的速度让一名新人（无论是销售、SDR、市场人员还是售后客户成功）具备“半个资深员工”的生产力，使其能透彻理解公司业务，推动客户转化并赋能终端用户。 ```mermaid flowchart TD accTitle: 知识沉淀与人员赋能飞轮 accDescr: 展示全员会议录音如何沉淀至内部大脑并反哺业务人员的完整闭环 subgraph Input["全域数据捕获"] A["全员日常会议与沟通"] B["Granola 等自动转录工具"] end subgraph Brain["内部大脑 Shrek"] C["知识沉淀池 / 沼泽看板"] D["企业价值观与北极星指标校准"] end subgraph Enablement["双向赋能交付"] E["被动查询：调用全球同行案例"] F["主动预警：日程联动与 ROI 话术"] end A --> B --> C C --> D D --> E D --> F ``` ### `35:49` 内部公司大脑“史莱克”（Shrek）与全方位知识沉淀 为了攻克快速赋能这一难题，Rogo 打造了一套具有魔力色彩的内部工具系统。所有进入 Rogo 的员工在第一天都会被告知：公司内部发生的每一次对话都会被记录并自动汇总到公司大脑中。这并非“老大哥在看着你”式的严密监控，而是因为所有人都在使用 Granola 等转录工具来强化个人记忆力，并让分散的知识能够复合式沉淀。 工程师们幽默地将这个内部大脑命名为“史莱克”（Shrek），系统中甚至有一个看板专门展示每个人当下正在攻关的“泥潭”（Swamp）。Shrek 深度集成了 Rogo 的所有底层系统，并被预先输入了公司的核心目标、企业价值观、客户交付标准以及“北极星指标”。这使得它能够以“主动与被动并存”的方式运转： - **被动查询**：如果一名员工正在向某家公募股票基金交付方案，即便从未与驻扎在亚太区（APAC）的同事见过面，也能一键调取对方三周前服务同类客户的完整对话记录，快速弄清对方曾处理过何种数据以及哪些用例行之有效；当遇到大机构需要评估模型路由（Model Routing）能节省多少成本时，系统也会立刻组织好话术与价值主张。 - **主动预判**：系统会自动扫描日历，例如提示“Patrick，我看到你周四要与一家私人信贷（Private Credit）机构开会。这里是该机构需要了解的全部背景信息、最具共鸣的实际案例，以及类似过往客户最看重的 ROI 衡量指标”。 ### `38:00` 吸引顶级人才的说辞：应用层 AI 与金融催化剂 当被问及如何将行业中最抢手、最具影响力的顶级人才挖到 Rogo 而非其他炙手可热的 AI 团队时，Gabe 总结了他面向候选人的核心宣讲逻辑。他坚信 AI 将彻底颠覆世界，但最激动人心、最富创造力的产品与工程工作必将发生在“应用层 AI”（Applied AI），因为这里正是 AI 与真实人类社会发生物理交汇的界面。 在所有行业中，金融扮演着驱动人类一切进步与创新的催化剂角色。资本配置（Capital Allocation）处于所有企业、所有创意乃至所有经济体融资链条的最上游；一旦能让资本配置的效率提升，整个世界的运转将被彻底超频。Gabe 强调，Rogo 是目前垂直赛道中最有希望“击中球门”的领头羊，团队的目标不是做一个 1000 亿美元市值的公司，而是打造一家市值达到 5000 亿美元、彻底重塑资本市场的巨头。面对极具深度与复杂度的技术难题，Rogo 组建了一支极其聪明、充满饥饿感、好奇且极低自负（Low Ego）的顶尖团队。 ### `40:00` 降低资本市场摩擦：让每一笔创新获得资金支持 现代资本市场的历史并不算漫长，仅有三四百年，但历史证明，随着市场流动性提高与摩擦减少，其对人类社会的正向效益会呈现倍数级增长。虽然金融从业者谈论“金融造福世界”容易被外界误解为私募巨头的自我粉饰，但回顾现代金融的起源即可明晰其本质： - **经典历史映射**：当年 J.P. 摩根（J. Pierpont Morgan）正是通过将欧洲雄厚的资本与当时作为新兴市场的美国本土企业家相连接，才得以资助铁路网建设与底层基建，最终让美国崛起为全球经济超级大国。其核心就在于中介机构让敢于承担风险的资本配置者与具有创新精神的创业家实现了高效对接。 - **巨大的未满足需求**：如今无论是美国本土还是全球范围，依然有大量实体经济无法触达资本市场。无数新兴国家的创业者无法为自己的创意融资或发债；即使在美国本土，也有约 30 万家中小企业甚至根本不知道高盛是做什么的，而摩根大通这类大机构也受制于人力成本，无法服务体量过小的企业。若能通过 AI 大幅加速企业家、创始人及个人接入资本市场的速率，全社会的创新飞轮将被极大推快。 ### `41:20` 创业者的“嚼玻璃”日常：在剧烈起伏中淬炼管理心智 在与其他垂直 AI 领域的优秀创始人（例如法律科技独角兽 Harvey 的创始人 Winston）交流时，Gabe 感触最深的是以极高速度增长背后所必须付出的“攻击性”以及日复一日“嚼玻璃”（Chewing Glass）般的痛苦。作为垂直领域的开拓者，必须在面临无数日常琐事全盘崩塌的同时，对行业长远的总体有效市场（TAM）与可服务市场（SAM）怀有无可动摇的坚定信念： - **如 Harvey 创始人 Winston 那样的定力**：在时刻承受上百处“皮肉创伤”的同时，依旧能完全无视杂音，全身心聚焦于三年后的终局状态，并只牢抓通往终局最重要的两件事。 - **高昂的创业学费**：Gabe 坦言自己在疫情期间工作，甚至没在传统办公室里待过，他和联合创始人 John 经常自嘲创业过程就像在上极其昂贵的商学院。由于缺乏经验，他们在招人、解雇、指导、管理、反馈机制以及方向制定上把所有能犯的错误都犯了一遍。 - **所谓“嚼玻璃”的具象场景**： - 核心员工离职、团队遭遇留存挑战； - 苦苦跟进挖了 6 个月的候选人最终拒绝了 Offer； - 投入巨大资源开发的某项功能上线后被证明完全是决策失误，让人顿感愚蠢； - 融资过程中连续被 40 家投资机构拒之门外。 - **情绪的极端振荡与升华**：创业就像坐过山车，在峰顶时感觉拥有全世界，在低谷时又觉得一切走向毁灭。但随着旅程继续，波峰越来越高，波谷虽然也更深，可当回头看三个月前的至暗时刻时，会发现自己“现在睡着觉都能轻松搞定它”。关键在于学会调节情绪，将情感转化为推动业务向前的动能，而不是让其演变为分心的阻碍。 ### `43:00` 红杉 Pat Grady 的逻辑与极致的进攻性（Aggression） 当探讨初创公司常挂在嘴边的“进攻性”（Aggression）时，Gabe 分享了红杉合伙人 Pat Grady 在其董事会上的犀利提问。当时 Rogo 团队刚刚汇报了一份在招聘、商业营收与产品进度上极为激进的下一年度规划，但 Pat Grady 用极简且无懈可击的两条逻辑前提彻底重构了视角： ``` 前提一：在接下来的 18 个月内，金融行业里的每一家机构都会对是否采购 AI 做出明确决定。 前提二：无论 Rogo 是否出现，他们都一定会做出采购决策。 结论：唯一重要的事情，就是不惜一切代价以最快速度席卷闪击市场（Blitz the market），确保你在现场。 ``` 基于这套逻辑，Pat 追问 Gabe：“考虑到这个现实，你认为你现在拿出来的计划足够有进攻性吗？”Gabe 当时的回答是“不够”。他意识到自己内心的潜意识其实在“手软”。作为由顶级风投支持的高成长企业，真正的策略是去主动拉宽结果分布的“两端尾部”（Expand the tails of the distribution）——哪怕为了追求成为千亿美元巨头的概率提升 20%，导致企业彻底失败的风险随之上升 30%，创始人也必须全盘接受并有勇气同时推高这两端的极端概率。 ### `45:00` 40 家顶级风投的拒绝信与被误解的金融市场 在寻求 A 轮融资之前，Rogo 的股东名册（Cap table）上并没有任何明星投资机构。早期 Pre-seed 和 Seed 轮投资人、BoxGroup 的 David Tisch 为 Gabe 引荐了约 40 位全明星级别的风险投资人，包括红杉资本（Sequoia）、凯鹏华盈（Kleiner Perkins）、基准资本（Benchmark）以及 Thrive Capital 等。Gabe 原本满怀期待，认为自己历经两年半的打磨，终于迎来了引入蓝筹资本、借助品牌黑洞效应吸纳顶尖人才与客户的历史机遇。然而，这 40 家机构全数拒绝了他。更残酷的是，这种拒绝往往伴随着极高深度的互动——投资人看了 BP 后觉得不错，约见 Gabe 聊上一小时，随后邀请他参加投委会（IC）、共进晚餐、甚至周末再度深入交流，但最终却冷冰冰地回复“我们不投了”。Gabe 形容这就像被 40 个自己深爱过的女友接连甩掉，因为在那个阶段，这种拒绝纯粹是对创始人个人的否定。 在这群人之后大约一个月，Keith Raboy 找到了 Gabe，尖锐地质疑道：“这根本不算一个具有逆向思维的赌注（contrarian bet），这不就是金融领域的 Harvey 吗？我为什么要投？”Gabe 当即反击：“如果这不是逆向押注，为什么你身边所有聪明的风投朋友都觉得这是个馊主意，并且全都不相信我？”Gabe 总结了当时风投看不懂 Rogo 的三大根源： 1. **严重低估金融市场的 TAM（潜在市场规模）**：旧金山湾区的风投界对金融缺乏底层直觉，因为像 Bloomberg、FactSet、S&P、PitchBook、Morningstar 以及 ION Group 这些利润丰厚的金融信息巨头，历史上基本都不是硅谷风投扶持出来的。 2. **早期产品确实很烂**：投资人本身就在金融行业工作，他们亲自试用 Rogo 时，回答有一半都是错的，于是轻率断定“这东西永远行不通”。他们忽视了模型指数级演进的速度，只要熬过 6 到 18 个月，技术代差就会完全抚平。 3. **怀疑创始人的破局能力**：当时投资人还没有足够的证据来确认，Gabe 是否具备持续“嚼碎玻璃”（chew glass）并在混乱中快速迭代求生的坚韧素质。 后来 Rogo 之所以能将硅谷和华尔街的顶级资方集齐，正是因为在后续的每一轮融资中，Gabe 都会一次次兑现那些曾经被资方断言“绝不可能达成”的里程碑。尽管过程充满波折——核心员工曾出走，看似敲定的大客户也曾半途蒸发——但团队展现出的极端动态应变能力，彻底赢得了资方的信任。 --- ### `47:41` 竞争壁垒与收购 6 家 AI 初创公司的技术拼图 金融垂直领域之所以缺乏真正有实力的竞争者，核心卡点在于极其严苛的获客与分发渠道（Distribution）。这本质上是一个如何与传统金融机构建立信任、交付高价值的人际问题，其难度甚至超过了单纯的工程与产品研发。与此同时，该领域的工程标准又高得惊人。Rogo 之所以能破局，得益于早期招聘了一批曾在高盛（Goldman Sachs）、花旗（Citi）、富瑞（Jefferies）、阿波罗（Apollo）、战神资产（Ares）和黑石（Blackstone）工作过的金融界精英。他们野心勃勃、极度聪慧，且具备毫无架子的空杯心态，愿意放下身段去“啃硬骨头”。 在技术团队的结构演进上，Gabe 指出单纯写代码的技术门槛正在降低，领域专长（Domain expertise）、产品直觉以及像总经理（GM）一样的全局业务思维变得愈发关键；而在底层系统扩容方面，又依然需要硬核粗粝的纯工程实力。为了补齐兼具商业直觉与顶尖工程能力的复合型人才，Rogo 采取了极度激进的策略：**直接先后收购了 6 家涉足金融 AI 的初创公司**。 ```mermaid flowchart LR accTitle: Rogo Talent Acquisition Engine accDescr: Flowchart showing how Rogo integrates acquired founder talent and Wall Street veterans into product and engineering. subgraph S1["人才输入来源"] A["华尔街资深从业者<br/>(高盛/黑石/阿波罗等)"] B["被收购的 6 家金融 AI 初创公司<br/>(经历过挫折的 Founder)"] end subgraph S2["核心复合能力"] C["深刻的金融业务直觉与信任感"] D["脑洞大开的产品架构能力 (Galaxy-brained)"] E["硬核工程与极速执行力 (GM Mindset)"] end subgraph S3["组织产出"] F["高壁垒金融 AI 基础设施平台"] end A --> C B --> D B --> E C --> F D --> F E --> F ``` 这些经历过创业挫折的创始人既拥有宏大的产品视野和对用户交互（UX）的敏锐洞察，又具备深厚的代码功底，成为驱动 Rogo 持续做出正确产品决策的中流砥柱。 --- ### `50:22` 金融业的创新者窘境与 Rogo 内部的“傻瓜帽”机制 过去一二十年里，高阶金融服务业与资本分配者的日子过得极为滋润。这是一个准入门槛极高的行业，尤其是在一级市场（Private Markets），只要募集了一期又一期的基金，其固有的商业惯性就让业务极难受挫。然而，这种安逸正在被彻底打破——金融行业正面临数十年来第一次真正的“创新者窘境”（Innovator's Dilemma）。未来将有成百上千家由 AI 原生构建的颠覆者、新型投资机构和投资银行，直接向传统老牌巨头的商业模式发起猛攻。 对于“AI 原生”（AI-Native）的定义，Gabe 认为这并非一句空洞口号，而是意味着**对 AI 极度痴迷（AI-pilled），敢于打破一切既定框架，没有任何业务模块被视作不可动摇的圣牛，坚定地将这种外星般的底层技术全盘织入组织的每一个角落**。 在 Rogo 内部，为了防止对 AI 的拥抱沦为一时走过场的口号，Gabe 建立了一套强硬且具有戏剧性的考核惩罚机制： - **全员月度 AI 工具使用率追踪**：系统每月会自动生成一份详细报告，统计每位员工使用内部自研工具及外部采购 AI 产品的时长与频次。 - **部门红黑榜公开处刑**：每个部门内部都会严格排名前 5 名的重度使用者（Power users）和后 5 名的低频使用者，并将榜单公开发布在公司各处。 - **佩戴“傻瓜帽”（Dunce Cap）**：每个部门排名垫底的员工，其头像会被打印出来并贴上一顶傻瓜帽，张贴在办公室显眼位置。随着公司规模扩大、新员工不熟悉 Gabe 的幽默风格，不少人感到相当惊吓，但这种近乎残酷的激励机制确实极大地驱动了全员对 AI 工具的实际应用。 --- ### `53:09` 传统金融掌门人必须反思的生死之问 如果一位执掌大型投资机构或投行的掌门人希望平稳度过这场技术海啸，Gabe 建议他们必须立刻向自己提出两个尖锐的核心问题： 1. **企业价值（Enterprise Value）的载体置换**： 如果说今天一家投资机构或投行 90% 的企业价值都依附在人才身上——依靠顶级合伙人和银行家的个人能力维系客户、寻找项目并创造收益；那么若明确知道在 10 年之内，**全球顶尖投资机构 90% 的企业价值将不再依附于自然人，而是沉淀在软件、私有数据和自主运行的系统中**，你今天应该启动什么？答案必然是：全力以赴将顶级精英潜意识与大脑中的暗默知识，系统性地抽离并固化为你所拥有、能自主运转的数智大脑。 2. **拆解交易全生命周期中的绝对壁垒**： 逐一剖析从寻找标的、尽调分析、撮合谈判到退出管理的完整 Deal 链条。诚实地叩问自己：在哪个具体环节你拥有别人绝不可能拥有的专有数据、关键人脉或不可替代的语境信息？如果你只是单纯自以为比别人看书更多、对行业更懂、脑子更灵光，那么这些所谓的“优势”很快就会被 AI 彻底抹平消融。 在整个金融业中，今天看起来最具有远见卓识的先锋，在两年前往往被同僚视为“彻头彻尾的疯子”。他们早在两年前就极力主张录制并转录工作中的每一场对话，将海量非结构化的决策过程实时灌入专属的“公司大脑”（Company Brain），甚至尝试为自己打造数字分身（Digital Clone），以此在复杂的商业博弈中推演最优策略。 ### `55:00` 投行的“第37步时刻”与私募市场的终局不确定性 在探讨金融机构对 AI 的认知转变时，Gabe 提到早在两年前，行业中就出现了极少数极具远见的领袖，例如 Moelis 的联合创始人 John Momtazee。当时多数人对 AI 还毫无概念，甚至纳闷“Patrick 到底在胡扯什么”，但 Momtazee 就已经敏锐地预见到未来：他明确提出希望为公司里最顶尖的银行家打造“数字分身”（Digital Clones）——这些系统能代替他们去开会、代表他们发言、深度理解他们的思考框架并吞吐所有上下文，最终目标是让投行里的每一个初级员工都能直接创造营收。这就如同围棋界的“第37步时刻”（Move 37 moment），少数先知先觉者突然顿悟到 AI 的深远变革力将远远超出常人的想象。 面对 Rogo 商业未来的终极不确定性，Gabe 坦言最难预判的是“私募市场（Private Markets）到底会以多快的速度被重塑”。回顾金融史，资产类别之所以能提升交易量与流动性，核心驱动力往往在于“标准化”，因为标准化大幅降低了追踪和交易门槛；然而私募市场充斥着海量的非结构化数据，长期以来对标准化免疫。虽然 AI 原则上能解决这一问题，但最终能否出现监管或某种市场外力，真正打破壁垒、推动透明度和流动性的跃升，依然不完全在软件公司的掌控之内。 另一个巨大的未知数在于：在人类核心关系中究竟还能保留多少超额收益（Alpha）。在小微盘的并购（Micro-cap M&A）场景中，一个倾注20年心血的小企业主在转手毕生心血时，必须看着接手者的眼睛握手建立信任，这一流程几乎不可能被完全自动化。但在由专业赞助商控股的企业（Sponsor-owned businesses）、二级市场（Secondaries）、私人信贷（Private Credit）以及 GP/LP 二级市场交易中，全流程自动化完全可行。最大的变数恰恰是那条漫长的长尾：无数面临代际传承更替的中小企业主，究竟要过多久才会习惯“点击一个按钮就能卖掉公司”，而不是非得当面握手交心。 ```mermaid flowchart TD accTitle: 私募市场不同资产类别的 AI 自动化潜力对比 accDescr: 对比关系驱动的长尾中小企业并购与数据驱动的机构化资产交易在 AI 自动化进程上的差异 subgraph HumanCentric["强人际信任驱动（难以自动化）"] A["微盘企业并购 (Micro-cap M&A)"] B["经营数十年的小企业主"] C["代际交接与握手信任关系"] A --- B --- C end subgraph SystemDriven["结构化与机构化（高度可自动化）"] D["赞助商持股企业交易 (Sponsor-owned)"] E["二级市场与私人信贷 (Secondaries & Private Credit)"] F["GP/LP 二级市场份额流转"] D --- E --- F end HumanCentric -->|"核心壁垒"| G["长尾企业主的信任机制与代际转变"] SystemDriven -->|"AI 标准化赋能"| H["数据结构化与高流动性自动交易"] ``` ### `56:46` 融资心法：拒绝自欺欺人，用确定性路线图兑现小概率胜率 针对年轻初创企业创始人如何与私募市场/风险投资人打交道，Gabe 分享了自己的融资哲学与避坑指南。他直言自己的风格极其直接坦诚，会毫无保留地告诉投资人自己最焦虑什么、目标要去往何方，但同时必须对公司的终局状态有着近乎偏执的坚定信念。融资本质上是一场“重复交互与建立认知”（Reps and relationships）的长期博弈。那些在 C 轮或 D 轮突然杀出领投的机构，往往在种子轮、A 轮、B 轮就已经接触过并连续 Pass 了多次；每一次拒绝都伴随不同的理由，但他们借此积累了创始人与业务在整个周期中兑现承诺的大量纵向数据。 在给创业者的忠告中，Gabe 强烈反对被滥用的“打肿脸充胖子”（Fake it till you make it）。虽然创业需要敢于展现霸气与气场，但这绝不意味着去伪装成一个不是自己的人。如果一家公司只有 5% 的概率成为千亿美元规模的巨头，创始人完全没必要硬装成有 100% 的确定性；真正专业且有说服力的姿态是：拿出一张逻辑严密、路径清晰的战略路线图，向投资人拆解“我们如何才能实现那 5% 的可能”，并展现出只要这些前提条件一一落地，就必定能登顶千亿市值的绝对自信。 - **透明直接的沟通风格**：坦诚披露业务短板与潜在风险，但在终局愿景（End state）上绝不退缩让步。 - **长期数据沉淀与关系迭代**：后期的领投方往往是早期一路观察、拒绝过你多次却见证你持续交付成果的熟人。 - **摒弃虚伪，拥抱理性野心**：无需假装自己毫无破绽，而是用可落地的战略路线将小概率转变为势在必行。 ### `58:30` 创业者的深层恐惧：如履薄冰的复利飞轮 当被问及为何自称是一个“内心极度偏执、极度缺乏安全感、深感恐惧的人”时，Gabe 动容地揭示了高强度创业的真实心理状态：每一天都仿佛行走在刀刃之上，时刻面临着千头万绪轰然坍塌的危机感。在他看来，做垂直 AI 软件这种业务，本质上是一场“复利动能”（Compounding Momentum）的残酷游戏。创始人的终极任务，就是把每一件微小的事情做到极致，让整台机器在下坡路上狂奔得更快一点。 真正让他夜不能寐的，是这种狂奔的动能突然停滞。一旦公司撞上路障偏离轨道，重新催化动能将付出极其昂贵的代价。打造一个具备自增强动能的机器难如登天，如果不能对可能击碎这一速度的任何潜在障碍保持神经质般的恐惧，就根本不可能穷尽一切手段提前做好风险对冲。 - **刀尖上的脆弱平衡**：每天都处在一千件事可能同时崩溃的边缘，容错率极低。 - **复利动能的建立与维持**：创业如同滚雪球狂奔下坡，极难加速却极易因磕绊而彻底熄火。 - **偏执驱动的极致防守**：正是因为深知重启势能的代价不可承受，才必须以神经质的警惕预防所有路障。 ### `59:37` 垂直 AI 的淘汰赛：成为人才与资本的黑洞 回顾在当前大模型时代的创业历程，Gabe 指出很多人低估了竞争的烈度。许多人总觉得在每一个垂直赛道都会自然诞生一个类似 Rogo 或 Harvey 的公司，尤其是在最后一公里布线极为复杂的企业服务领域。但现实是极其残酷的：虽然 AI 带来了空前强大的时代东风，但今天的企业执行门槛比以往任何时候都要高，所有人都在被迫卷入残酷的大联盟（NFL），面临严苛的“欢迎来到职业联赛”时刻。 最终能够存活并通吃的企业，必须在极短时间内蜕变为“吸纳人才、资本与品牌的超级黑洞”（Black holes for talent, capital, and brand）。在基础模型实力一日千里的背景下，没有顶尖人才团队根本无法跟上节奏，而组建这支梦之队的价格高昂到令人咂舌。如果一家企业不能迅速构建起黑洞效应来虹吸资源完成超高难度的执行，就极大概率会沦为底层前沿模型实验室（如 Anthropic、OpenAI）碾压下的牺牲品。 ```mermaid flowchart LR accTitle: 垂直 AI 赢家的“黑洞飞轮”模型 accDescr: 展示垂直 AI 企业如何通过资本、人才与品牌黑洞构建执行力，避免被底层模型实验室吞噬 subgraph ResourceIngestion["超级黑洞效应"] A["资本集聚 (Capital)"] B["顶尖人才虹吸 (Elite Talent)"] C["品牌信誉沉淀 (Brand)"] end subgraph HighBarExecution["极致工程与业务落地"] D["高成本工程交付"] E["最后一公里复杂布线"] F["超敏捷响应迭代"] end subgraph EndgameDefense["终局壁垒"] G["跨过职业联赛级竞争门槛"] H["防止沦为前沿实验室的牺牲品 (Roadkill)"] end A & B & C --> D & E & F D & E & F --> G G --> H H -.->|"巩固壁垒再虹吸"| A ``` ### `01:00:46` 善意的一体两面：无条件的狂热信任与极致严苛的托底 在访谈标志性的收尾问题——“他人为你做过最友善的一件事是什么？”中，Gabe 动情地回顾了父母两种截然不同却深刻塑造了他生命的善意。他的母亲代表的是纯粹无条件的爱与信任：无论 Gabe 做了什么混账事、在学校考砸还是在创业初期遭遇 40 家机构连续拒绝，母亲永远坚定地认为他是世界上最聪明、最不可思议的天才。尽管这种盲目在客观上是极度非理性的，但正是这种来自至亲至善的非理性自信，在每一次跌入谷底的暗黑时刻给了他撑下去的底气。 相比之下，父亲的善意则表现为一种严肃而厚重的“极致标准”。面对失败，极其自律、高标准的父亲往往气愤到难以直视 Gabe，完全无法理解儿子拥有如此优渥的机会为何还会搞砸；但他表达爱的方式，是强忍着不解与怒火，亲自搬来椅子坐下，逐字逐句、从头到尾陪着 Gabe 复盘剖析每一个犯错的细节，全力帮他纠错提升。Gabe 总结道，正是母亲无底线的托底自信与父亲不妥协的卓越标准，共同锻造了他直面千亿征途的抗压骨骼。 ## Notable quotes > `01:24` 未来每个机构的数据中心里都会装着一个国家规模的天才群体。 — Dario Amodei > `16:57` 对于 Anthropic 来说，做这些事情就像走在通往万亿美元营收的大道上，停在路边去弯腰捡一分钱硬币；但对我们而言，这正是构筑数百亿美元垂直帝国的核心支撑。 — Gabe Stengel > `38:15` 在严肃金融场景中，可审计性（Auditability）甚至比绝对准确率（Accuracy）更为重要。 — Gabe Stengel > `33:45` 我刚给客户发了一份 5 页的高质量分析材料。如果在过去，我必须和分析师来来回回沟通修改整整三天；而现在，我自己只花了 10 分钟就全部做完了。 — Gabe Stengel 引述投行董事总经理 > `43:00` 在接下来的 18 个月内，金融行业里的每一家机构都会对是否采购 AI 做出明确决定。无论 Rogo 是否出现，他们都一定会做出采购决策。唯一重要的事情，就是不惜一切代价以最快速度席卷闪击市场（Blitz the market），确保你在现场。 — Pat Grady > `45:00` 如果这不是逆向押注，为什么你身边所有聪明的风投朋友都觉得这是个馊主意，并且全都不相信我？ — Gabe Stengel > `50:22` 对 AI 极度痴迷（AI-pilled），敢于打破一切既定框架，没有任何业务模块被视作不可动摇的圣牛，坚定地将这种外星般的底层技术全盘织入组织的每一个角落。 — Gabe Stengel > `53:09` 全球顶尖投资机构 90% 的企业价值将不再依附于自然人，而是沉淀在软件、私有数据和自主运行的系统中。 — Gabe Stengel > `58:30` 做垂直 AI 软件这种业务，本质上是一场复利动能（Compounding Momentum）的残酷游戏。创始人的终极任务，就是把每一件微小的事情做到极致，让整台机器在下坡路上狂奔得更快一点。 — Gabe Stengel > `59:37` 吸纳人才、资本与品牌的超级黑洞（Black holes for talent, capital, and brand）。 — Gabe Stengel ## Mentioned Gabe Stengel, Rogo, John, Pat Grady, Keith Raboy, David Tisch, Winston, John Momtazee, Max Levchin, Dario Amodei, Lee Sedol, J. Pierpont Morgan, OpenAI, Anthropic, Sequoia Capital, Kleiner Perkins, Benchmark, Thrive Capital, BoxGroup, Goldman Sachs, Morgan Stanley, JPMorgan Chase, Bank of America, Citi, Jefferies, Moelis, KKR, Blackstone, Apollo, Ares Management, Millennium Management, Citadel, Jane Street, Bloomberg, FactSet, S&P Capital IQ, PitchBook, Morningstar, ION Group, Harvey, Cursor, Factory, Claude Code, Granola, Robinhood, Rocket Mortgage, Opendoor, Affirm, Shrek, GPT-3, o1 Pro, Opus 4.5, AlphaGo ## Video description Gabe Stengel is the co-founder and CEO of Rogo, the AI platform for finance. He believes the best investors will spend the next several years reinventing their firms around AI, and Rogo is trying to build the infrastructure that allows them to do it. We discuss how Rogo evolved alongside the frontier models, why the last mile and the harness around the models matter so much, what happens when every portfolio manager can deploy thousands of agents against a problem, and how AI could transform the way capital is raised, assets are priced, and deals get done. We also cover which investing skills become more valuable as AI improves, the move from seat-based to outcome-based pricing, Rogo’s internal company brain called Shrek, the 40 investor rejections Gabe received before his Series A, why building in applied AI requires extraordinary aggression, and what it takes to become a black hole for talent and capital. Please enjoy my conversation with Gabe Stengel. For the full show notes, transcript, and links to mentioned content, check out the episode page ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠.&nbsp; ----- Become a Colossus member to get our quarterly print magazine and private audio experience, including exclusive profiles and early access to select episodes. Subscribe at ⁠colossus.com/subscribe⁠. ----- ⁠Ramp’s⁠ mission is to help companies manage their spend in a way that reduces expenses and frees up time for teams to work on more valuable projects. Go to⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ⁠ramp.com/invest⁠⁠ to sign up for free and get a $250 welcome bonus. ----- Trusted by thousands of businesses, ⁠Vanta⁠ continuously monitors your security posture and streamlines audits so you can win enterprise deals and build customer trust without the traditional overhead. Invest Like the Best listeners get a special offer of $1,000 off Vanta when you go to ⁠vanta.com/invest⁠.&nbsp; ----- WorkOS⁠ is the infrastructure B2B and AI-native companies use to sell to enterprise. It covers everything enterprise security requires: SSO, SCIM, RBAC, Audit Logs, AI governance, and more. Trusted by 2,000+ fast-growing companies, including OpenAI, Anthropic, Cursor, and Vercel. ----- Rogo is the AI platform for finance. They're building agents for Wall Street that are trained to understand how bankers and investors actually do work: from diligence and modeling, to turning analysis into deliverables. To learn more, visit rogo.ai/invest. ----- ⁠Ridgeline⁠ has built a complete, real-time, modern operating system for investment managers. It handles trading, portfolio management, compliance, customer reporting, and much more through an all-in-one real-time cloud platform. Visit⁠ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ridgeline.ai⁠. ----- Editing and post-production work for this episode was provided by The Podcast Consultant. Timestamps: (00:00:00) Welcome to Invest Like The Best (00:01:24) The Model Eras of Rogo (00:04:15) Why the Last Mile Mattered (00:06:14) Who’s Using Rogo Right Now (00:10:55) Which Investor Skills Still Matter (00:12:23) Inside Rogo's Data Stack (00:14:33) Competing With the Frontier Labs (00:16:57) Why the Harness Matters Most (00:18:05) What Makes a Vertical AI Winner (00:22:00) How Firms Buy AI Software (00:23:43) From Seats to Outcome Pricing (00:31:19) Auditability Beats Accuracy (00:32:55) Scaling Enterprise Sales Fast (00:34:48) Meet Shrek, Rogo's Company Brain (00:35:49) The Pitch to Great Talent (00:39:37) What Chewing Glass Feels Like (00:42:18) Forty Investor Rejections (00:47:41) Finance's Innovator's Dilemma (00:50:22) Questions Every Firm Should Ask (00:53:09) What Remains Most Uncertain (00:56:46) Becoming a Black Hole for Talent (00:57:56) The Kindest Thing [Source](https://traffic.megaphone.fm/CLS7922860814.mp3)

[1:24] Dario Amodei: 未来每个机构的数据中心里都会装着一个国家规模的天才群体。 [16:57] Gabe Stengel: 对于 Anthropic 来说，做这些事情就像走在通往万亿美元营收的大道上，停在路边去弯腰捡一分钱硬币；但对我们而言，这正是构筑数百亿美元垂直帝国的核心支撑。 [38:15] Gabe Stengel: 在严肃金融场景中，可审计性（Auditability）甚至比绝对准确率（Accuracy）更为重要。 [33:45] Gabe Stengel 引述投行董事总经理: 我刚给客户发了一份 5 页的高质量分析材料。如果在过去，我必须和分析师来来回回沟通修改整整三天；而现在，我自己只花了 10 分钟就全部做完了。 [43:00] Pat Grady: 在接下来的 18 个月内，金融行业里的每一家机构都会对是否采购 AI 做出明确决定。无论 Rogo 是否出现，他们都一定会做出采购决策。唯一重要的事情，就是不惜一切代价以最快速度席卷闪击市场（Blitz the market），确保你在现场。 [45:00] Gabe Stengel: 如果这不是逆向押注，为什么你身边所有聪明的风投朋友都觉得这是个馊主意，并且全都不相信我？ [50:22] Gabe Stengel: 对 AI 极度痴迷（AI-pilled），敢于打破一切既定框架，没有任何业务模块被视作不可动摇的圣牛，坚定地将这种外星般的底层技术全盘织入组织的每一个角落。 [53:09] Gabe Stengel: 全球顶尖投资机构 90% 的企业价值将不再依附于自然人，而是沉淀在软件、私有数据和自主运行的系统中。 [58:30] Gabe Stengel: 做垂直 AI 软件这种业务，本质上是一场复利动能（Compounding Momentum）的残酷游戏。创始人的终极任务，就是把每一件微小的事情做到极致，让整台机器在下坡路上狂奔得更快一点。 [59:37] Gabe Stengel: 吸纳人才、资本与品牌的超级黑洞（Black holes for talent, capital, and brand）。

[1:24] Gabe Stengel: nothing worked at all [4:15] Gabe Stengel: spidey sense [10:00] Gabe Stengel: student of these guys [10:00] Gabe Stengel: great chef [16:57] Gabe Stengel: stopping on the side of the road to pick up a penny [32:55] Gabe Stengel: deal team of one [34:48] Gabe Stengel: it's actually more important to be auditable than it is to be accurate [35:00] Gabe Stengel: The same way that there's going to be individual investors that make horrible decisions and you need to go in and see what went wrong. Was there an incorrect data input? Did someone lie to them? What was going on, you're going to need to do the same thing with agents. [35:49] Gabe Stengel: Finance is a domain that is the catalyst for all human progress and innovation. And capital allocation is upstream of the financing of every company, every idea, every economy. And so if you can make that more efficient, you can supercharge the world. And we're the category leading player who is the best shot on goal and not just be the hundred billion dollar business to do it, but the five hundred billion dollar business that completely transforms capital markets. [39:37] Gabe Stengel: Didn't know how to hire, didn't know how to fire, didn't know how to mentor, didn't know how to manage, didn't know how to give feedback, didn't know how to set direction. And a lot of the people problems that arise with scaling quickly feel like eating glass to me anyway. And so when people quit, when you have retention issues, when you spend six months recruiting a candidate and they don't join, that's chewing glass. That bug that comes out and makes you feel like an idiot for spending all that time and capacity on something that was the wrong call, chewing glass. When you get rejected by 40 investors in a row before you're able to raise capital, chewing glass. [55:00] Gabe Stengel: You need to believe that that 5% likelihood that you can be a $100 billion business is likely. And you don't need to pretend it's 100% likely, but you should be able to delineate with very clear roadmap and strategy how it is possible that you can become a hundred billion dollar business, and then have confidence that if those things play out, you will be. [56:46] Gabe Stengel: AI is an amazing tailwind, but the execution bar is higher than it's ever been to compete. And everyone is getting pulled into the big leagues and is having their welcome to the NFL moment. You have to move faster and be stronger and be more resilient, agile than you ever expected. And having the right team is more important than ever. And the right team these days is extremely, extremely expensive.

[0:00] Ramp is the only platform built to make your finance team leaner, faster, and better, saving businesses 5% annually on average so you can stay focused on growth. Ramp just opened for business in the United Kingdom, so if you're running a business in the UK, you can now use Ramp's AI-powered finance platform to manage cards, expenses, bills, approvals, and accounting all in one place. I run my business on Ramp, and so should you. Learn more at ramp.com slash invest. Topic, Perplexity, and Vercel all have something in common. They all use WorkOS. And here's why. To achieve enterprise adoption at scale, you have to deliver on core capabilities like SSO, SCIM, RBAC, and audit logs. That's where WorkOS comes in. Instead of spending months building these mission-critical capabilities yourself, you can just use WorkOS APIs to gain all of them on day zero. That's why so many of the top AI teams you hear about already run on WorkOS. fastest way to become enterprise ready and stay focused on what matters most, your product. Visit WorkOS.com to get started. you and i have talked many times about this basic question that i'll start with over the last couple of years you are effectively trying to build investing super intelligence tools to help investors do their job much faster better cheaper easier higher quality But it's starting to feel like Wow, we're really eating a lot of the core functions that even a very smart analyst or even portfolio manager was doing a couple of years ago. How do you think about that trajectory as you've seen it and lived it so far and where it's going over the next two years? I think two years is actually easier to reason about than 10 years or 20 years. Because in two years, the best investors are going to be figuring out how to reinvent their own firms and reinvent themselves. And if you look at what happened to market making and quant trading, Jaden Street took 15 years to build the dominant franchise. And the world's best investors today are going to spend the next two to five years figuring out how to integrate AI into what they do. And Dario has the great line about everyone's going to have a data center full of geniuses or a country full of geniuses in the data center. What would Goldman do? What would Millennium do? What would Citadel do if they had a country full of geniuses show up? It would probably take them a while to figure out how to change the way they work, how to take advantage of that, how to integrate it into their system. I think figuring out how to apply AI into the investment lifecycle. is the biggest challenge over the next five years for every great investor. There's been many companies... on this trajectory where The product... was cognition is very famous for like literally their ads now say remember devon like it's good now so lots of now clearly great companies with great products had a stage of of an AI business where the product stunk and now it's excellent. If you think about the couple increases in capability that we've seen just from the raw models, could you do the same thing for the eras of Rogo and its product like You pick how many eras it is. I don't know how you frame it up, but like what it could do at each level up to and including today. yeah i actually tried starting rogo two times before we got started so i In high school, I had a friend whose dad was an investment banker who wanted an app. for trying to track the basically equity exchange rate of two public companies as they were merging. And so we tried using really old AI techniques to do that. Terrible. And then in college, before GB3 came out, we published a paper on AI assistance for econometrics and financial econometrics. And we tried commercializing at the time and nothing worked at all. And then when we actually started the business, it was when GB3 came out pre-Chat GBT. And so all the early days of Rogo, it was clear how kind of magical it was. You could demo things that were cool. Nothing worked at all. I mean, I would say since things actually started to work, The eras are very tied to the model eras, right? It was 01 Pro and then probably Opus 4.5. 01 Pro is the first time you got enough reliability where it was a good search tool at the very least. You could say, help me calculate this financial metric for this business over the last 12 quarters. And it could do it reliably enough where it wasn't so annoying that you would just do it yourself. And then with Opus 4-5 and the last year, the end of 2025, end of the beginning of this year, The models just became capable of basically anything a junior investment professional or junior banker was doing as long as you gave it the right instructions and context. I mean, I think there was a first movers disadvantage for a lot of applied AI companies because you thought you knew where the world was going and you wanted to build a product for it. But the models weren't quite there. And so people would try to go, this is terrible. This is garbage. You have a disadvantage. For us, we saw that too. But now what we've seen is Well, if you were right about the end state and where the models were going and you were building towards that, when they get there, it's magical. And for me, a great product is all the feedback we get every day. People all day are saying, hey, this is transforming the way I [5:00] I'm saving hundreds of hours a month. I am doing things I never could have done before. And so I'm smarter as a result and I'm able to make better decisions and, And it's delightful. And I like using it. And, you know, it brings me joy in my day to day because the UX and the attention to detail and the craftsmanship is so obviously built for me and who I am. And so that's been the best part of the product. And you would attribute that to. You took seriously all the compliance, regulatory workflow, last mile hookup stuff, and then the models became good enough. And all of a sudden that was super valuable. Well, there's also small details of understanding how someone within one of these firms works and building for it. I'll give you an example. We make it so easy for a managing director at a bank to email a markup of a deck, which is how they're typically... doing these workflows anyway, except sending it to an analyst and sending it to our AI analyst over email and then returning that markup on 20 minutes as opposed to two days. And at the same time, alert the junior analyst on the deal what's happening and show them the full auditability of all the little markups that were made in case they want to weigh in. And that whole UX, that whole flow just makes it so much easier for this financial professional who is not logged into a computer in 10 years, but does have an iPad where they know how to mark these things up to actually And there are these small details of how you build a product that's great for a specific end user that you only know if you have the kind of spidey sense for what the job is. What is like the bleeding edge of what it can do that impresses you the most? What kinds of jobs? The coolest things that we're working on is taking these innovations like Maltbook. Imagine if every PM at a hedge fund had 10,000 agents that were just kind of fraternizing, talking about ideas, reading through the notes, pontificating, and then at the end of 24 hours of debate, just gave you one idea. And the reason you're able to do that is because investors are happy to pay $50,000 for one really good idea, whereas there's very few other domains where you can expand that many tokens just for one simple insight. What we're seeing right now, though, is that the models are smarter than Anyone I know, anyone I spend time with, and it's plumbing connected to your context, informative about your thesis, tell it the way that you work and try and integrate it into what you do. And so building out all the plumbing to actually collect that data, collect that context is what is cutting edge to me. It's always interesting to me for a product like this that, You're opinionated about what it should be used to do, but in some sense, people can be creative with how they use it. So you get to sort of revealed how people want to use it. if I adopted a God's eye view of... It's Monday morning here in New York City. Lots of roger users are probably fired up and using it right now. If I could somehow see into every instance of the product being used, What would I see? Who are the people? What are the predominant use cases? How varied are they? Give us a sense of how it's being used right now as we record. Even though in so many ways, I think public equities is the best application of AI because all the data is available. And so it's just about being as smart as possible. That's a very brute force framing. Our early users in ICP and kind of core market is actually what I would describe as deal makers or people that are transacting, who are buying companies, selling companies, and helping coordinate transactions. And so a lot of what we do is both make people smarter, but actually do the deal making. How do you prepare a data room? How do you unpack a data room? How do you coordinate the call with the third parties to discuss the data room? How do you go through all of the initial steps through closing of a deal? And so if you took a bird's eye view of all the folks using Rogo, I mean, it's people who are either on the sell side of a transaction or the thoughts and materials to help execute that full deal, whether it's putting things into a data room. This is the company's model. This is the PowerPoint that describes their customers. These are the answers to the DDQ questions on what customer concentration is. Or it's all the agents on the other side of that that are tearing through the data and mapping it to the firm's investment philosophy to say, oh, great. You know, is it lower than the concentration risk profile that we would want for this fund, too? And then the components of the system. R. people access it via all the classic channels you would access an ai tool email, chatbots, proactive alerts, and those kind of things. But then Rogo is actually in a lot of the behind the scenes systems of these firms, because when you're working on a deal, it's not just important for the human beings working on it, but you need a update your CRM. You need to update your portfolio and monitoring systems. You need to update the way that you distribute information to your LPs after the fact. And so half of our surface area is actually the underneath of the iceberg of interacting with these different systems of record based on what the humans are doing over the course of the deal. So dealmakers today, when do you think you'll be able to give the same answer for junior analyst at a public equity hedge fund or something like this. there's a process to their workflow [10:00] as well, but it's very, very different. You're right that it's interesting that my first intuition would be public markets are the best place to do this because there's so much data available. When do you think that transition happens? I think that for our business, we need to have all the requisite domain knowledge of what it takes to be a great public markets investor. I don't know what it takes. I've never done it. I haven't spent nearly as much time as I should have with the folks that are great at it. And we need to both hire out that domain expertise and then figure out based on it how to apply the systems that we have built to that market. I have extreme conviction in the fact that the underlying systems and tools and infrastructure we have built will be invaluable to that market. But now we need the great chef who can figure out how to piece it together and create that kind of end state product and that last mile delivery for public equities investors. I get pushed a lot by our board to think about expanding the ICP beyond just core banking. But the reality has been is there's been so much depth and tan in this dealmakers vertical. And then for my end state vision of actually being the full infrastructure for private markets where people can transact very effectively that is far more important to the dealmakers Whereas for public equities all that infrastructure all those exchanges already exist And so I like to serve them because I want to serve the most sophisticated smartest users who have inordinate amounts of knowledge on the companies they track and the industries they follow. And I'd like to make them even smarter because that sounds super cool. But I can build a huge, huge business just concentrating where I am today. Interesting. So one takeaway from that would be a lot of the opportunity to build an AI business in a vertical is somewhere where there's lots of plumbing that's not yet built. Yes. And then applying your data on top of that. Exactly. I mean, part of the reason private markets are so attractive is because it's all done by humans, the coordination, the standardization, looking into things and the actual transacting. Whereas public equity is a lot of it has been automated. But based on what you know, what skills do you think investment professionals, broadly speaking, public and private, should think about being or becoming more valuable as time progresses and which skills become kind of obvious the skills that become less valuable? What the hell are we going to do? I think can do the soup to nuts diligence and IC memo and objection handling and all this kind of stuff Like that a big part of a job for at least a junior person in the investing world So what skills do you think people will still matter a lot in a couple of years I mean I want to preface it all with I worked in finance for two years and so I am a student of these guys just as much as I fascinated by the technology and want to figure out how to use it But the world's best investors have a way of figuring out what matters and exercising their own judgment across a range of topics. And I think jury is still out on whether or not that is something that I can eventually replace. When move 37 happened and Lisa Dahl saw something I could never see, if that starts to happen in public equities, yeah, it's going to really change what matters. And if that really changes all of how that works, I mean, I think the core skill set is folks who can go out and gather data and inputs into their model that no one else will have. If you can spend time in the field, if you can speak to experts, if you can develop a relationship graph of folks who can inform your model, maybe you're not the one that needs to calculate what your move. Thirty seven would be for a great public equities investment, but you can actually feed your model with data that no one else has to. Thank you and gives you instant up proof of trust without all the manual work so you can close deals faster and with less friction Customers report a 526 return on investment and more than 16,000 companies use Vanta, including Ramp, Harvey, and Snowflake. Get started with $1,000 off at vanta.com slash invest. Ridgeline is the first end-to-end system of record with embedded AI for investment management firms running portfolio accounting, reconciliation, reporting, trading, and compliance on one unified platform. Firms are moving off legacy technology and onto Ridgeline because of how far ahead Ridgeline's AI features are compared to anything else in investment management software. I've been hearing from a lot of investment managers about AI, and they fall roughly into two camps, with some unsure where to even start and others convinced they can build their own order management system over a weekend. The reality is that running an investment firm will always require governance, controls, and a single source of truth for your data. And no amount of AI enthusiasm changes that requirement. Ridgeline is built on exactly that foundation, which is why I believe that the firms that come out ahead in the AI era will be the ones running on Ridgeline's unified platform. If you're serious about your firm's AI strategy, Ridgeline should be part of that conversation. And you can request a demo at ridgeline.ai. [15:00] Maybe it's a good time to talk about, I think, about the assembly line, like if Rogo... puts out some really useful output. I want to learn about each component part of the system that leads to that output. And I'm most especially interested in the data that you yourself have and use and buy and build and whatever, and how you do that, how you think about data, and then how you think about model. And these are both questions, one that I'm curious about your specific business for. But I also think that there will be some business like Rogo built in basically every vertical. And so I'm curious what can be abstracted to other professional services or other verticals that are interesting as well that will fall to AI. progress. So talk us through specifically like the data and model piece and how that's evolved over time. The early days of Rogo was kind of this Roop Goldberg contraption where like you had 60 different model calls. A question comes in, you try to say. What companies is Patrick asking about? Now, what are their tickers? Now, how do I feed those tickers into an API called a Bloomberg or fax it or some internal data set? Now it comes back and I need to call a different model to pull it all together. And as the models get smarter, you kind of want to. be less prescriptive, less of a Rube Goldberg machine, and just think about what are the simplest, best, highest quality tools. And the same way that if you have the world's smartest human being starting here tomorrow, trying to be a great banker and investor, what are the tools that they would need that are not just intrinsic to being smart, right? What are the data tools? What are the ways of going out and gathering information? What are the ways of auditing its own work? And then what are the ways of presenting it and pushing it back into the systems that it needs? And so for us, we spend a lot of time thinking through what are all the data inputs or a great investor would need to actually do their job. Then we spend a lot of time thinking about, great, when you're doing that job, what are all the compliance and regulatory requirements to make sure that if someday some Delaware state, court judge makes AI inputs and research discoverable, you have actually done it all the right way such that you're not intermingling information because the reality of AI investment judgments and AI banker outputs is that you're going to be able to see the full lineage of how those things are created. And then we spend a lot of time benchmarking these models and creating different evals and data sets. So we can always decide what is the most performant? What is the cheapest from a token perspective? What is the lowest latency and route tasks to the appropriate type of models? I'm sure your favorite question is. How do you ultimately compete with Anthropic and OpenAI who view finance as a big one of the few categories that like you see them talking about and thinking about what can you do in the long run that's counter positioned against what they can or will do? Do you think? I mean, you want to build things that are perpendicular to what they want to build. And sometimes you might build a chatbot because that helps you go to market faster. But you should know there's a whole bunch of stuff underneath the surface that the labs are never going to build that we need to build for finance or for any other vertical. And when you think about financial services and capital markets, How many businesses are there that generate more than five or ten billion dollars by just going deep into those workflows into the data sets and how work is done? There's a huge amount of TAM and spend. on top of very messy, specific problems. And all of finance is a collection of different niches with different data sets, different definitions of good, different regulatory requirements. And we can get to $5 billion in revenue by going deep across those things and creating the systems, the systems of record that help manage them. That for Anthropic would kind of be like stopping on the side of the road to pick up a penny because they're on the pathway of trying to go from 100 billion in revenue to a trillion in revenue. And there's so much depth to these systems that actually need to be built beyond just intelligence. Is there a favorite example of that of some like pain in the ass thing that you've had to wire up last mile things? Yeah, I mean think about If you are ingesting MNPI because you are working on transactions or you're working on deals that have an effect on the market, the compliance requirements you have just in the auditability or what you can flag and how it feeds into the internal systems of an investment firm or a bank so that if you ever get audited or a regulator ever wants to see what you did, do you have all that plumbing in place that is pixel perfect? and you need to send data back and forth, you actually need some sort of data room, something that is compliant, safe, and secure, that coordinates everything. And ideally, it's not just some static kind of Dropbox folder, but something that's plugged into the way that you do work, your agents, your workflows. And I don't think OpenAaronthropic will ever want to build a data room business. And if you actually want to be the exchange for all of high finance and all of capital markets, you not just need to own the intelligence. You need to own the transaction venue, the communication venue, the workflows and all the data inputs that go into it. It's interesting. I asked about data and models, but actually it sounds like harness or infrastructure is probably the most important of the three. Think about the fundamental difference between Claude Code when it came out and Claude Cowork versus OpenAI and Chatsby T. The models were actually fairly similar, but the harness and the way that it was presented from Claude was far better, and it just allowed the models to exercise more of their long [20:00] running capabilities. And that's why they had a run up in usage and a huge amount of expansion. And it just shows the way that you harness these models is so, so important. And I think people underappreciate that intelligence, the reasons that humans are high agency and can do a lot is not just because we have high raw recall and IQ and knowledge, but there's all these different microservices in your brain. How do you put knowledge away? How do you retrieve it? How do you trigger things? Emotions are a way to trigger all these different microservices. that's why a great investor might have great judgment, is because they have good spidey sense for when they see this sort of thing in the market, it actually triggers the recall from this event that informs a creative decision. All those kind of small microservices are things that need to be built out. But if I force you to become an investor and your only goal is to invest in rogo-like businesses, like one of these businesses that's, let's say a vertical application that's wiring up the capabilities of AI to an industry, what features would you look for that would get you the most excited, either in the industry or in the founder, builder, and their approach based on what you've learned? A few things. One is the industry actually does have to have enough complexity and depth in the types of data the types of systems of record that people use the types of deployment models that you can spend a lot of effort solving those problems in order to have a wedge to solve everything else Because if it an industry that anyone can just walk into and sell the basic version of ChatGPT or co or co-pilot immediately, and you don't have to solve all these weird integrations, you're not going to have enough time to build all those things that are perpendicular to what you're doing. And so the industry itself needs to be adjacent enough to the core market. So that's one thing. The second thing I would look for is domain expertise from the team and the founders. And I don't have unique domain expertise, but I had enough to get started. And then I was so curious about finance and money and capital markets. And I grew up in New York and I was surrounded by people that all they could think about, all they could talk about was high finance. And I was fascinated by it and I wanted to learn about it. And then we assembled a team that was uniquely passionate about it too. And so I have over a hundred people that have spent time within investment banks or within investment firms across the world's best institutions. And so we can constantly take the models as they're released and harness them for finance. And our job is really to catch the change in the models and figure out how to apply it in these institutions. The final big thing I would look for is a business that willing to constantly reinvent the core product and constantly willing to slash it to nothing And I think anyone who their delivery method of their product is not something that they can fully cannibalize quickly like a terminal or like a very specific UX or interface is not going to be agile enough to constantly reinvent every six months when there's a step change. What's an example of that that you've done, like a tear down and rebuild? The story, the anecdote I'm most inspired by is Max Lenshin, who talks about how to firm they rebuild the fundamental ledger technology every year and they rebuild it for a few reasons. One is it's the most interesting engineering problem. And so all the engineers want to work on it. So it's a good way to retain talent, to teach engineers about the core fundamental business of a firm. But then number two, it's a good way to make sure that system doesn't ossify and it's constantly improving. And so we do that same exact thing for our harness and the core agentic system. We are constantly looking at it, realizing we're not even at a local minima, it would be impossible if we were at a local minima because the models are changing so quickly and we need to redo the whole thing. What's something that the models currently cannot do that if they could, would really change the nature of the product? Compaction. So it's, if you have a hundred conversations with a single agent how does it make sure that it actually remembering the right things and compacting its memory into an amount of tokens that it can use every time And it has enough coherence and context on who you are and what you care about to make it feel like it a true person that you speaking to that you know learns more and more about you That's a hard problem. And it compounds exponentially when you think about agents, not just as a one to one. Right now, almost all agents are one to one. You use ChatGPT individually. You use Copa individually. You use Gemini individually. as soon as these are actually things that consort with a lot of colleagues or in a Slack channel with 100 people have to work across an entire company. Now, the compaction problem just scaled exponentially because it's having conversations with 100 different people and needs to be able to coordinate across those things. And so being able to take all of that memory, all those interactions and actually lodge it into the mental model or brain of that agent so that it can be persistent so that it can actually maintain context over the course of a bunch of interactions. that's something that the models are not great at today. What do you think the major kinds of AI software businesses there are? So we've got companies like a cognition or a cursor or something that can grow unbelievably quickly. And I'm especially curious for you to compare this, like the old classification system for software companies. I'm curious how people buy rogo, what kind of category you would put it in? Is it usage-based? Is it seat-based? Is it something else? Like talk us through the, how people want to [25:00] buy this stuff, what the emerging models are for AI software business. We are a classic enterprise software business. We price per seat right now because our buyers are used to pricing per seat. They think of us in a similar category to... Bloomberg to FactSet to Capital IQ to PitchBook. And so we have to build a very human business. Every time you sign a deal, it requires an AE and a solutions architect or the sales engineer going in and shaking a lot of hands, explaining how it works, explaining how to integrate it. You can't sign one deal where the usage just rises a hundredfold. I look at how hard it was for Anthropic to sell to us very easy. And the amount that we pay them has risen exponentially without a human loop because it's a token consumption model. There's a lot of industries where riding the coattails of token consumption isn't going to work for enterprise sales. And we're one of those. And so it's actually pretty interesting because we have to build a go to market machine. five times faster than most enterprise sales organizations ever have to build. And so I do think there's the category that's just typical enterprise sales, but using AI models as a tailwind to build 100 times better products. And then there are the token brokers, the token consumption businesses, where you're selling in the parts of enterprise where they're used to buying usage based tools, cursor, factory, cloud code and others. And so you can go much further commercially with fewer people. What's your prediction for how or if that will change in finance? I think that every business needs to go through two different pricing revolutions. You need to move to some sort of usage based and then you need to move to some sort of outcome based. For me, if I can figure out a way to skip the token base, skip the usage base, simplify it for my users and just wait until I can say, hey, Patrick, what if I just charge you for every good investment idea I give you? Or what if I charge you for the quarterly report you send to LPs that I can do perfectly? Or what if I charge you for every SIM that you create as a banker? I would much rather get there than have to figure out some random way of trying to assign dollars per token. That is something that we're not going to quite agree on because you're going to spend $100,000 on tokens and say, well, I get $100,000 of value and I don't really know. But you know what the value is to you of a good investment idea because you can actually see how much money did I earn? Or you know what the value to you is if you can produce a sim if you're a bank because you know what you charge these firms to actually sell the business. Presumably, you can't change your seat price on the fly dynamically, at least with the same customer. So how do you deal with the problem of like, in some ways, misalignment with the customer for your business, where if you do a great job and they use the thing way more, which costs you a lot of money, they become a worse customer. The reality is, is we are 1% of the way into our product roadmap. 99% of the innovation for capital markets is in front of us. And so what matters is that we're a good partner. We're a good steward of their AI strategy, and they want to work with us in the future. It's so interesting to think about the shape of this in the future. You said you're 1% penetrated into the roadmap. Give us a sense of where you think this is all going from your product perspective, not industry-wide. If you have that much of the roadmap ahead of you still, describe that to us. Think about the percentage of all capital markets, workflows, investment workflows, investment banking, jobs that are still completely human rate limited and done by human intermediaries. It's kind of similar to other parts of financial services where 15 years ago, 20 years ago, Every mortgage that someone got, you went in, you spoke to a banker at a local branch. It felt like a very human decision. I'm buying a house. I'm taking out a loan. This is important. I need to speak to someone. No one ever thought that you wouldn't want a human in the loop for that. Now, 40 to 50% of mortgages are just delivered online by platforms like Rocket Mortgage. I think there's going to be a huge amount of innovation and how companies transact, how companies raise capital, how companies raise debt. I think it'll be easier than it's ever been in 10 or 20 years for someone who's a business owner or someone who works at a company to go online and click a button and try to raise capital the way that someone can go on to Robinhood and click a button and buy an equity. I think it'll take five minutes for KKR to figure out, can I sell this portfolio company to another sponsor? Not five months. I think you're going to be able to price the assets in an order of magnitude less And as a result, markets are going to be more transparent, more liquid, more efficient, and there's going to be a whole bunch more activity. I'm going to focus on the specific future of automated risk pricing. for lack of a better simple term. We're three years from now and everything you just said is true, where I can raise single digit millions of dollars of better equity, kind of like filling out an online form and the thing can just price the risk for me and give me an offer. It's like open door for like everything or something. What do you need to build? that you don't already have to enable that sort of capital market future? It's actually a very similar strategy to Bloomberg strategy. Bloomberg strategy was, I'm going to omit a lot of details here, but offer a little bit of data again in the door, build all the analytics and workflows on top that someone would need, and then provide the exchange and the communication platform where you can actually transact in a bunch of [30:00] of asset classes that before it was pretty opaque, Bloomberg Messenger. For us, it's use a little bit of AI to get in the door, build out the full workflows, go from co-pilot chatbot to full autopilot tool. So I can make sure that I'm 100% accurate on your IC memo or on the DD queues that you're doing. and then provide the communication channel between counterparties so that if I have agents that can autopilot do the work, I can actually transact for you. And the difference between what I need to build and Bloomberg Messenger is I don't need to build the communication channel for humans to transact. I need to build the communication channel for the agents to transact across these businesses, across these investment firms. And so if you think about. What is the actual infrastructure that needs to get built? I mean, think about what is the system that would allow a large private equity firm to feel comfortable having an agent negotiate a deal on its behalf, correspond with all the third party consultants in the transaction, the people doing the QOV, the legal advisors and so on, and then actually run an auction process where you have a bunch of sponsors providing bids. There's a huge amount of software to be built out. And I think sometimes we talk about that as looking like an exchange for a lot of these asset classes that are not standardized. Thank you. Tell us a little bit about the customer base today, like how much of it is giant banks versus investing firms. It's mostly large banks. And that's very simply because that was my background, right? I worked as an investment banker for just a handful of years doing buy side M&A coverage, which was super interesting. And so we started targeting the banks pretty early on. for a few simple reasons. One is the investment banks are kind of the distribution channel for the rest of finance. A lot of the folks who then end up as great investors started in their first two years as an analyst at Goldman in the TMT program or something of that nature. Two, they have the most seats by far. And so if you can land a bank like Bank of America, you can actually get in the hands of far, far, far more people than if you land the 10 best single portfolio manager public equities investors who each only have 10 investors. If I think about a Bank of America or something, everyone's kind of wondering how deep into the adoption curve are we for enterprises using AI? you have a biased sample because your customers are using robo and they're using it a lot. But give us a sense of where you think we are? It seems really hard to pin down a good answer. I would say the majority of firms are seeing a huge amount of individual productivity. And they're trying to figure out how do we parlay that individual productivity into firm productivity that we can measure. Speak to any individual banker at a bank we're deployed with. Life's great. Oh, they're like, I'm 100 times more efficient than I used to be. Right. You'll speak to an MD who will say, Gabe, I sent five pages to a client that before I would have had to go back and forth with an analyst on over three days to create. And I made it in 10 minutes myself. And these are bankers who haven't done any sort of analysis in 25 years. They haven't actually opened an Excel file in 20 years, and they're able to do it themselves. The problem is, where is that flowing through? Are you winning more deals? Are you actually transacting more? Are you servicing a part of the market that you haven't seen? And this is where it becomes not just an individual productivity tool problem, but a firm strategy problem. Like, what's your plan? Do you want to use this thing to cut costs? Do you want to use this thing to enter parts of the market that before didn't make sense to serve? You can look at a bank like JP Morgan. JP Morgan just announced that they're going to try and do a lot more M&A work for SMBs for parts of the market that before they didn't think it made sense to go out and serve because the deal fees were probably too small. And so you needed too many people to staff them. Well, now if you have a banker that can be a deal team of one, well, maybe you can enter parts of the market that before just made no sense. So in some sense, the bottleneck at some point will be the creativity of the customer, right? Like you can provision unlimited capability. And you're soon going to be just relying on them figuring out the answer. It's them figuring out what they want to do. Right. If you had 100 great investors start here tomorrow working for you. How would you channel that productivity? It would take you a while to figure out what's the structure. Do they all work on different things? How much money do I give each of them? What do we attack? Two years ago, the most obvious question, and this would have been about accuracy and people to use the word hallucinations, which seems to have dropped out of the conversation. And I can't remember the last time that someone said hallucinations to me. What can you teach us about? That problem. ensuring accuracy where accuracy matters a lot to the decimal. What's the nature of that these days? It's still super important. I think it's actually more important to be auditable than it is to be accurate. And obviously those two things are conflated. But what's really important is that I give you an answer. You know how to use it. And if it's not accurate and you don't know how to check it and it's hard to see where it came from, you can't use it at all, whether it's accurate or not, because you don't trust it. If it's accurate most of the time, but even when it's not, it's very easy to see the assumptions that went in where the data was pulled from. It's still actionable and it still saves you time. And then increasingly, as these things go from co-pilot tools that you're just using for information retrieval to autopilots where you're trusting them not just to gather the information, but to execute on it, to have agency and actually make an investment decision or send [35:00] email. You need to have the full confidence that if you were to go back in and see why it made the decision, you would be able to understand why, because you're going to need to debug it. And the same way that there's going to be individual investors that make horrible decisions and you need to go in and see what went wrong. Was there an incorrect data input? Did someone lie to them? what was going on, you're going to need to do the same thing with agents. And then especially in parts of capital markets that have regulators that look at these things and need to make sure there's no foul play. If you can't explain why you made a decision and the data went in, that's not going to fly. You were talking before about how you have to sell like a normal enterprise sales organization would. You have to grow or can grow many multiples faster than the fastest growing enterprise SaaS companies of the last era. How do you solve that problem? You're very limited by the speed of humans to some degree in enterprise sales. How do you hire enough people fast enough? How do you think about being able to grow at the right rate when you don't have the anthropic cognition, API usage growth. It's so easy for them to grow 10x. It's much harder for you. How do you solve that? The core problem we need to solve is how fast can you make a human being productive as a half person? Salesperson, but anyone else as a marketer, as an SDR, as a post salesperson, how quickly can they understand our business and help push us forward and push customers forward and help our end users? And so enablement and training people and constantly retraining people is the fundamental problem that we and I assume other fast growing enterprise startups have to deal with. So do you like use AI to build tools? Oh, the internal tools we have are kind of magical. First off, every internal conversation that happens at Rogo is recorded. When anyone starts, we say, hey, just FYI, Patrick, you're always being recorded. It's always being filtered into the company brain. This is not in a kind of big brother situation. It's just everyone you're going to speak to is going to have granola or some eating transcription running because they need to use it and they need to have excellent recall and they need to compound the knowledge that they have. And as a result, we just have this huge reservoir of information. And then we have all of these tools that people can use on top to say, oh, you know what? We're deploying with this sort of public equities firm in this sort of market. have we ever served this kind of data before? Are there use cases that might be helpful? And you can pull in the conversation that a peer of your head three weeks ago, and you never even met that person because they're stationed in APAC. Being able to sponge all that information in and then get it out to people when they need it is the core problem of enablement. So maybe describe the internal brain. It's called Shrek for some reason, because my engineers thought it would be hilarious to call it Shrek. There's a dashboard where you can see the swamp of everything that people are working on at any given time. But it's connected to all of our different systems. It is very prescriptive about what it knows our company goals are. What are our values? What are the things we want to deliver to clients? What are our North Star metrics? And so it can shape every answer and deliverable that way. And it's both proactive and reactive. Someone can go in and say, I'm trying to get up to speed on how I should talk about model routing and how I should think about the value prop there for a very large institution and what the savings will be. And it can pull out all that information for you. But it can also say, hey, Patrick, I see on your calendar on Thursday, you're meeting with this sort of private credit firm. Here's all the information you should know, all of the use cases that will resonate. And then all of the types of ROI metrics that firms we've worked with in the past would want to hear. So what's the sales pitch to talent? let's say there's somebody that, if you landed them tomorrow would be transformative because they're so high quality or well-known or whatever. What is the pitch to them to come work at Rogo versus go somewhere else that's exciting right now? Always depends on the person's motivation. So it's hard to give a generic pitch. But my pitch for Rogo today for Talon is... AI is going to completely transform the world. The place it is going to be the most interesting is applied AI because that's where AI intersects with humanity. And so the companies that dictate how AI intersects with humans and touches humans are going to do the most interesting, creative engineering and product work in the world. Finance is a domain. That is the catalyst for all human progress and innovation. And capital allocation is upstream of the financing of every company, every idea, every economy. And so if you can make that more efficient, you can supercharge the world. And we're the category leading player who is the best shot on gold and not just be the hundred billion dollar business to do it, but the five hundred billion dollar business that completely transforms capital markets. And there's such a depth and a complexity and an amount of interesting problems that's so exciting. And we have a killer group of people that is super smart, hungry, curious, and low ego that's going to do it. Good fucking pitch. Enjoy. Glad I invested. Say more about this capital markets piece. Historically, as markets get more efficient and liquid, we're going to be able to get Their positive impact, in my opinion, grows a lot. You can chart this through market history, which isn't that long, three, four hundred years of like proper markets. [40:00] What do you think is possible? Like, where might this be going and why do you believe that? that creating more or less friction, I guess, in capital markets can be so powerful. you're always at risk of sounding like the, billionaire private equity guy saying that private equity is good for the world when you talk about how finance is good for the world. But I like to think about the origins of high finance. And when you think about a business like JP Morgan, Some of the origins are J. Pierpont Morgan helping connect European investors with the entrepreneurs in an emerging market. the United States to finance railroads and the infrastructure build out. And everything that allowed the U.S. to be a juggernaut economy. And that happened because there were intermediaries who helped connect folks who were risk takers and capital allocators with the folks who were entrepreneurial and wanted to innovate. And that was something that had a profoundly. Great effect on the world. And now think about all the parts of the economy, all the parts of the U.S. domestically, but also internationally that can't tap into capital markets. Every emerging nation where you would struggle to raise capital, to finance your idea, to raise debt. And then the 300,000 American businesses that couldn't even tell you what Goldman Sachs does. And JP Morgan doesn't have the time to go out and work with them because the business is too small. If you're able to speed up the rate at which entrepreneurs and company founders and individuals can tap into capital markets, you can accelerate all innovation. What are you learning from your peers that are building companies kind of shaped like yours, but in other categories? I am learning... how much aggression it takes to grow this quickly. I am learning how much chewing glass it is on a day to day basis. And how much conviction you need to have in the long term, Tam and Sam, to make sure that you don't fuss over all the things that are going completely wrong every single day. And, you know, I'll call someone like Winston at Harvey. and Winston's ability... to just not worry about the hundred flesh wounds that are inflicted on him at any given time. And just think about the kind of end state goal of where he's going to be three years from now. And the only two things that matter to get there. is pretty amazing. What is the glass like and what is the aggression like? What does that mean? I worked during COVID, so I didn't even get to see what an office looked like. John, my co-founder, and I used to joke that it's a very expensive business school education because we were just doing everything wrong. Didn't know how to hire, didn't know how to fire, didn't know how to mentor, didn't know how to manage, didn't know how to give feedback, didn't know how to set direction. And a lot of the people problems that arise with scaling quickly feel like eating glass to me anyway. And so when people quit, when you have retention issues, when you spend six months recruiting a candidate and they don't join, that's chewing glass. that comes out and makes you feel like an idiot for spending all that time and capacity on something that was the wrong call, chewing glass. When you get rejected by 40 investors in a row before you're able to raise capital, chewing glass. And my experience of startup building is it's like a roller coaster where you have to feel the extreme highs and feel the extreme lows. And I'm a super emotional guy. And I try not to let the team feel it, but I will feel on top of the world at the high and like everything is cataclysmic at the low. But then if I look back at the journey, the lows get lower, the highs get much higher. And I look back three months ago at the low I was dealing with or the high I was dealing with. I was like, I could do that in my sleep now. And I think it's just about modulating those things and channeling the emotion to push the business forward, but not letting it distract you. What about aggression? Like that seems kind of like a trope. You've interviewed Pat Grady. Pat Grady was at our board meeting on Wednesday, and we presented what was an extremely aggressive plan for next year in terms of hiring goals, commercial goals, product goals. And one of the reasons I love Pat is because he boils everything down into like two bullet points and it's logically infallible. And, you know, he's just like premise one, premise two. This is the result. He goes, Gabe, well, if everyone in finance is going to make a buying decision on AI in the next 18 months and they're definitely going to buy something no matter what, even if you're not there, then the only thing that matters is that you can blitz the market as quickly as possible to make sure that you are there. Given that, do you think this plan is aggressive enough or no? The answer was no. And the reason it wasn't aggressive enough is because I was being soft. And I think the reality is you need to be so, so, so aggressive and underwrite all that risk and know the game that you're playing. And my goal is if that interback business is to increase the tails of the distribution. It's fine if it gets 30 percent more likelihood that I fail. If the odds that I become a hundred billion dollar company also increased by 20 percent. But you actually have to be okay with raising both of those tails at the same time. What's the most emotional low that you've faced? when we were raising our series [45:00] guy We didn't have any star investors in our cap table yet. But there was a great investor, David Tisch at Box Group, who is an early pre-seed seed investor, who introduced me to a bunch of all-time greats for the Series X. And I thought, wow, I've seen what it does to get a blue chip investor in your cap table, what it does for recruiting talent. becoming more of a black hole for brand and customers and so on. This is finally the opportunity we're going to have to do that. We've been building for two and a half years. And David introduced us to 40 investors and I met with everybody. I met with Sequoia, Kleiner, Benchmark, everybody. And I'm like, 40 people passed and it wasn't just like, You've got the email with the deck and it's not exciting. It was like, oh, this is interesting. Let me meet Gabe. Oh, I kind of like Gabe. Let me spend an hour with him. Oh, Gabe, come to IC. Oh, Gabe, let's go to dinner. Oh, Gabe, come in for the weekend afterwards. You know what? We're going to pass. And it's so personal because at that stage, it has nothing to do with anything but you. Right. It's like it's like being broken up with by 40 girlfriends who like every time you fell in love and then every time they said not for you. And that was actually Thrive to the series. I thrive spent so much time with me. We went to dinner with Avery and Vince, fell in love with them and the firm because they were awesome. And then crushing blow. And luckily, Keith Raboy came basically a month after everyone else had rejected us. And Keith was like, Dave, this isn't a contrarian bet. It's basically just Harvey for finance. Why would I do it? And I said, Keith, if it's not contrarian, why did every single one of your friends just say it was a bad idea and not believe in me? Why do you think they didn't believe? For a lot of reasons, I think people underappreciated the TAM in finance, which was silly. But it's partly because I think SF has less intuition for finance because they didn't fund Ion Group or Bloomberg or S&P or Faxed or PitchBook, Morningstar and so on. And so there actually haven't been great venture backed businesses in this market. And so there's no intuition for the kind of contours of the market and how large it is. Two is the product was terrible. And so every investor said, oh, I work in finance. Let me use it. Is it going to transform what I do? And they tried it and it was wrong half the time. They go, this is never going to work. And it's like, guys. You know, we're on the exponential. It's going to work in six months or 12 months or 18 months or whatever it is. I'm going to figure it out. And the final reason was people didn't think I could figure it out. People didn't have enough data points of watching me chew glass and watching me figure out what the next iteration would be. What do you think changed after that? Because now it's a who's who of captive. Well, because I met all these guys very early on and I met them at every single round. And so every time I said, hey, we're going to do this. And they said, yeah, there's no way. And then every time we would do it. And by the way, a lot of the times it would manifest in a different way. you would actually lose the key employee you needed or that actually excellent customer that you thought you were gonna land totally Dissipated. But we kept figuring out what to do and kept navigating the market. And I think when you're investing in a market like today, especially at Applied AI, where it's so turbulent, so ambiguous, you need to underwrite the founders being able to be extremely dynamic. Why do you think there aren't more credible... financial services competitors. Distribution is really hard to crack. I think it's far more a people problem on building trust. and delivering value and working with these institutions than just an engineering and product problem. But then there's an enormously high engineering and product burden, too. the standard to execute to really crack this market, I think is pretty high. And I was just so lucky that a number of the early people I hired were ex-finance and just killers. And we just hired folks from Internetwork who worked at Goldman or City or Jeffries or Apollo or Aries or Blackstone. And so we had a team of people who were diehard, ambitious, curious, smart, and then just good humans, low ego, humble, young enough to be open-minded and to want to eat class I'm also really curious how you've dealt with technical talent and what matters in a technical person on your team and how that has evolved. One intuition might be the value of domain expertise from the technical person has gone up as the ease of execution has evolved. Changed a lot. You don't need to be super technical to write good code or less so than in the past. So what is the shape of the engineering or technical team look like over time? I'm asking this again because I'm curious about thinking about other companies that want to tackle this and whatever their industry might be. So I would say there's definitely problems where domain expertise. is increasing in value and product intuition and having more of a GM like mindset versus an engineer like mindset is super, super important. There are also parts of our product surface area where it's like you just need raw, gritty engineering talent because you're scaling things up so aggressively. But if I look at the folks on our team who have been so excellent, a lot of them are former founders. They're folks who started businesses. persevered, ate glass, had a lot of product intuition, figured out how to channel it. And then the business may be petered out. And to the point you made, there's been a lot of companies trying to tackle financial AI. There's been a lot of really smart, product-minded [50:00] but engineering minded, domain experts who have tried to tackle finance AI. And we've acquired six different fledgling financial AI startups. And so those former founders who can be galaxy brained about the future of the product and navigate a course on what the UX should be, but also have engineering chops. And so they can constantly make the right decisions is super important. You said earlier to me before we were recording that this is the first time that we have like a major innovators dilemma in this category. Can you describe what you mean by that? Yeah, well, I think for a lot of investment firms, a lot of high finance financial services, the last. 10, 20 years have been pretty good. You can make a lot of money being a capital allocator, being an investor. It's a very hard industry to enter. And then especially if you're in private markets, your businesses have some natural momentum because when you raise a fund, another fund, another fund, it's hard to mess up the business after that. And I'm sure there's 100 people who have started funds who know. It's 100 times harder than I just described. But to be fair, there hasn't been a moment and a shock to the market where every investment firm, every bank is saying, oh, wow, I need to completely rethink what I'm doing. And now there's going to be an opportunity for hundreds of AI native disruptors, AI native investment firms, AI native investment banks to attack my business model. And I need to figure out what I'm going to do now. And there has not been an innovator's dilemma for private equity firms, for hedge funds, for investment banks in the long time. AI Native, what does that term mean? mean to you? Like, what is the definition of that? In some ways, I think it just means willing to constantly reinvent everything you're doing and being so AI pilled that you don't worry about what's possible or what might seem completely far-fetched, but you were just charting a trajectory towards integrating this Alien fundamental technology into everything that you do and it's showing up in every part of the business and there's no part of the business that you hold sacred that is immune to being revolutionized. How do you do that? Like so you're a leader in a company that's very AI pilled that want to make sure you want to do everything you guys said in Rogo. And yet I'm sure there are areas that you're unhappy with the status of how much AI is used to do X, Y, or Z. how do you do it as a leader? How do you make sure that your company keeps doing this as a practice and habit versus like a one time. So I'll give you a very simple thing, which is every month. I have a report that gets sent to me that shows me everyone at the company, how much are they using the various different AI tools we have, both the ones we've procured, the ones we've built internally, all of these different things. And I have a stack ranking within every division of the top five power users and the bottom five users. And we post that everywhere. And the bottom user in each division gets a printout with a dunce cap and we post around the office. And it's hilarious, but people know it's coming. And by the way, as we get bigger and fewer people know me and know it's kind of a joke and I'm funny, they're actually terrified. And I hope they're not terrified. And I hope that changes. But it's a strong incentive function. to actually make sure you're using these things. If you were to list questions, let's say there's someone sitting down that runs one of these firms that's been a great business, hard fought, but a great business. And these businesses tend to be quite simple from an organizational and technical standpoint. It's mostly humans, right? There's not a lot of overhead in lots of these businesses. Maybe they buy a lot of data or something. What questions would you encourage them to ask of themselves, of their business, and to stand the best chance of navigating this transition effectively. If you knew for sure that right now 90% of your enterprise value is in your people, Your best investors, your best bankers are the people that bring in deals, bring in revenue. And actually, that's what accrues enterprise value. And in 10 years, the world's best investment firms, best banks, will have 90% of their enterprise value, not in people, but in software and data and systems, what would you start doing? You would start trying to figure out how do you take all of what lives in the latent minds of your best people and putting it into a system that you own and operate autonomously. So that's a big one. The second thing I would think about is. unpack every part of the deal life cycle and In each one, try and chart where do you think you are invaluable or you have data or you have domain expertise that no one else has. And then be diligent about saying, OK, do I actually have something that no one else has in this market? And is it a relationship? Is it context that no one else has? Or do I just think I'm smarter and better read on the subject area in the market, in which case AI is going to obviate that? Have you seen anyone be the most cutting edge exemplar of this attitude in a big firm? Is there a favorite example of a person that's just... Oh, yeah, there's a racing of the there's it's the folks that today look the most prescient sounded. crazy batshit two years ago. They were the guys that came in and said, "We need to record every conversation. You're going to be able to pipe in this conversation directly into my company brain, and there's going to be a digital clone of me. And then it's going to spit out the game theory on exactly [55:00] the investment should look like. And two years ago, everyone listened to this kind of guys and were like, What the heck is Patrick talking about? One example is there's a co-founder of firm Mollis, John Montazzi, who was just prescient about where it was going. And he wanted digital clones of all their best bankers. He wanted systems that could basically show up to calls, speak on his behalf, know how he thinks, then ingest that context. of... every junior in that bank to be revenue generating. And so there are a number of folks who had their kind of move 37 moment where they realized, wow, this is going to be so much more profound than anyone's expecting. What to you feels the most uncertain about the future of your business? I would say how quickly private markets actually do transform. If you think about why different types of asset classes have increased in transaction volume and have increased in liquidity, often it has to do with standardization because it gets easier than to track those assets and trade them. Private markets have been immune to standardization because there's so much unstructured data. AI should fix that. That said, is there going to be a regulatory or a market force that forces some additional the kind of transparency and liquidity you can have in private markets. That's a little bit out of my control. The other thing is it's still unclear to me how much alpha will be left in the human relationships. Right. I talk about that micro cap M&A. It's very hard to imagine that if you're a small business owner and you've been building a business for 20 years and you want to make sure that if you hand that business off to someone else, that you trust them, that you can shake their hand, that that can be fully automated. But for a lot of sponsor-owned businesses or things like secondaries or private credit or GPLP secondaries, I do think it can be fully automated. But how long it takes for the long tail of... All these small businesses, all these medium sized businesses that have to deal with generational turnover for them to get comfortable clicking a button to sell their business as opposed to shaking the hand to someone. That's a little up in there to me. If you had several new young founders here with us and they were curious about how to navigate and interface with private markets investors, how would you have to navigate and interface with private markets investors? What would you teach them? What would you tell them to do, to not do? My style for fundraising might not be everyone's style. I'm super direct. I'm super transparent about what I'm worried about, where I want to go. But then you have to be very headstrong on that end state. And then I would say it's about... reps and relationships and the folks who come out of nowhere and lead the series C or the series D are the folks that I met at the seed and then the A and then the B and the C and they passed every time for all sorts of reasons but they gather a lot more data on me and the business Anything that you would encourage people to not do? I think there's a lot of the fake it till you make it. And you need to have the bravado and the confidence in what you're doing, even if you're not fully confident. And I'm a deeply paranoid, deeply insecure, deeply scared person. But you need to put on the face and say that you are confident about where you're going, even in the moments where, you know, it's the ebb and flow, it's the ebb and flow. And I think People can misinterpret that sometimes as like they need to pretend there's something they're not. I think that's not true. You need to believe that that 5% likelihood that you can be in $100 billion business is likely. And you don't need to pretend it's 100% likely, but you should be able to delineate with very clear roadmap and strategy, how it is possible that you can become a hundred billion dollar business and then have confidence that if those things play out, you will be. Why are you scared and insecure? I'm so paranoid about everything that can go wrong. Every day it feels like you're on the knife's edge of a thousand things collapsing. And I think the reality of this sort of business building is that it's a game of compounding momentum. How do you do every small thing to just race a little bit faster downhill? And I'm just scared the momentum will stop. Or you hit a roadblock and then you go off course. And then you need to recatalyze momentum. And it's so clear to me how hard it is to actually... build a machine that gathers momentum. So if there's any stumbling block that halts that velocity, if I wasn't constantly petrified of those moments, I wouldn't be doing everything under the sun to prepare for them. What have we missed? Like what that you've learned about building a company like this in this era? where everything feels like a jump ball. Like it feels like there's going to be a Rogo or a Harvey or whatever for every place that there can be. Especially where the last mile wiring is hard and it's not just going to be anthropic. It's one company to rule them all or open AI. What else have we missed that you think is really important to the experience so far of building the business? I think that there are going to be businesses that solve all these problems. But the businesses that do it are going to become black holes for talent and capital and brand. And they're going to be able to siphon in the resources to actually execute. [60:00] Because AI is an amazing tailwind, but the execution bar is higher than it's ever been to to compete. And everyone is getting pulled into the big leagues and is having their welcome to the NFL moment. You have to move faster and be stronger and be more resilient, agile than you ever expected. And having the right team is more important than ever. And the right team these days is extremely, extremely expensive. And so if you can't figure out what is the strategy to become a black hole for talent and capital as quickly as possible, I do think you are far more at risk of being roadkill of a lab or a company that can. When I do these, I always ask my favorite question last. What is the kindest thing that anyone's ever done for you? I benefited so much by having... parents who were enormously kind and generous and selfless. but it showed up in such different ways for my mother versus my father. And... My mother's version of kindness was no matter what I did, I was amazing and smart and could do no wrong, even though growing up. That was absolutely not the case, but she instilled in me the confidence to believe in myself. Even when you were in those low trajectories where it felt, Like everything was going to go sideways or anything. Right. 40 knows in a row. Idiot failing out, flunking out, whatever it is. No matter what, she acted like I was maybe the smartest person on earth. That was irrational. But you need some of that irrational confidence that comes from just undying love. My dad was very different. My dad, if I came home and I had done something wrong, was seething, could barely look at me, couldn't understand it. Right. Like he was someone who was so disciplined, so good, had such high standards. And so his version of kindness was figuring out how does he understand who I am and why I am failing at this thing and then help me. He would sit down and go over every detail with me, even though he sometimes just couldn't even understand why I had the opportunities I had and I couldn't take advantage of them. and take the time to make me better while staying true to his principles and his standards of excellence and his definitions of good, too. You're building a fascinating business. It's been so cool to watch it get better. Thanks for doing this with me. Thank you. more at colossus.com slash subscribe. Patrick O'Shaughnessy is the CEO of Positive Sum. All opinions expressed by Patrick and podcast guests are solely their own opinions and do not reflect the opinion of Positive Sum. This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. Clients of Positive Sum may maintain positions in the securities discussed in this podcast. To learn more, visit psum.vc. Outro Music. You know how small advantages compound over time? That's true in investing and just as true in how you run your company. Your spending system is your capital allocation strategy. Ramp makes it smarter by default. Better data, better decisions, better economics over time. See how at ramp.com slash invest. As your business grows, Vanta scales with you, automating compliance and giving you a single source of truth for security and risk. Learn more at vanta.com slash invest. The best AI and software companies from OpenAI to Cursor to Perplexity use WorkOS to become enterprise ready overnight. not in months. Visit WorkOS.com to skip the unglamorous infrastructure work and focus on your product. Ridgeline is redefining asset management technology as a true partner, not just a software vendor. They've helped firms 5x in scale, enabling faster growth, smarter operations, and a competitive edge. Visit RidgelineApps.com to see what they can unlock for your firm. Every investment firm is unique and generic AI doesn't understand your process. Rogo does. It's an AI platform built specifically for Wall Street, connected to your real outputs. Check them out at rogo.ai slash invest.