1mind's Kvarfordt Links AI Pilot Failures to Bad Data, Not Software

1mind的Kvarfordt将AI试点失败归咎于不良数据而非软件 企业AI项目之所以瓦解，是因为营收负责人在修复底层工作流程以及设定与销售速率挂钩的严格终止日期之前，就购买了工具。 Jonathan Kvarfordt解释称，大多数企业AI项目的失败并非源于技术缺陷，而是由于技术优先的心态、数据孤岛以及薄弱的变革管理。营收负责人不应盲目追逐泛化的AI战略，而必须利用销售速率等式将实施方案锚定于业务成果，并将工作流程细化到具体的人工作务和数据负载。他提出了OAR框架（优化、扩展、重塑，即Optimize, Amplify, Reinvent），以根据组织的成熟度和运营就绪程度指导AI嵌入的深度。为确保切实的问责制，他敦促领导者建立单一的可量化董事会指标，并执行与领先转化指标挂钩的严格30至90天试点终止期限。 进入市场（GTM）团队往往在明确所需解决的实际业务问题之前，就频繁询问该购买哪款AI工具。 (0:04) 尽管多项调查显示企业AI采用率很高，但实证分析表明，在2025年上半年仅有7.6%的机构（2025年下半年为24%）真正将AI投入运营，而不仅是停留在基础席位许可阶段。 (1:47) AI项目的失败源于三大非技术原因：技术优先的心态、数据不成熟以及薄弱的变革管理。 (4:21) 机构应当构建运用AI的业务战略，而非孤立的AI战略，并将落地实施锚定于销售速率等式。 (4:21) OAR框架将AI部署划分为优化（加速传统流程）、扩展（跨步骤拓展范围）和重塑（围绕买方偏好重构全流程）。 (7:51)

Jonathan Kvarfordt argues that enterprise agentic AI pilots routinely collapse because revenue leaders purchase software before diagnosing underlying operational bottlenecks, data architectures, or human workflows. Sustainable return on investment requires anchoring technology within core business metrics, matching operational readiness to the appropriate level of change, and enforcing disciplined pilot evaluation criteria. The Illusion of Enterprise AI Adoption (1:47) In 2025, executive mandates sparked board-level urgency across go-to-market teams, propelled by corporate directives like Shopify's internal memo, Duolingo's hiring freeze on roles AI could perform, and Salesforce CEO Marc Benioff's assertion that AI handles 30% to 50% of internal work. Kvarfordt observes that this barrage of announcements created intense anxiety among revenue leaders pressured to demonstrate immediate returns. However, broad enterprise adoption remains largely an illusion. While external reports like McKinsey's study claimed an 88% adoption rate, Momentum's empirical analysis of customer conversations revealed that actual operationalization reached only 7.6% in the first half of 2025 and 24.0% in the second half. Kvarfordt contends that most organizations merely distributed chatbot seat licenses without redesigning underlying revenue workflows, suffering from excessive tool prescriptions without adequate organizational diagnosis. Root Causes and Strategic Reverse-Engineering (4:21) AI initiatives falter primarily due to three organizational breakdowns: a tech-first mindset that prioritizes software purchases over operational diagnostics, data immaturity marked by disconnected functional silos and poor security guardrails, and deficient change management for human adoption. To illustrate, Kvarfordt cites two peer companies deploying AI SDR platforms like Artisan and 11x; one succeeded while the other failed entirely due to an inaccurate Ideal Customer Profile rather than software defects. To reverse this trend, leaders must avoid isolated AI strategies and instead build business strategies enabled by AI. Kvarfordt recommends reverse-engineering target goals through the sales velocity equation—multiplying opportunities, average contract value, and win rate, divided by sales cycle length. Revenue leaders must isolate which specific lever is lagging and distinguish whether the shortfall stems from rep execution, strategic positioning, product-market fit, or systems infrastructure before applying automation. Navigating the OAR Spectrum and Organizational Friction (7:51) Go-to-market AI implementations span an 'OAR' continuum—Optimize, Amplify, and Reinvent—where the entry point depends on organizational maturity and risk appetite. Optimization accelerates existing legacy processes with minimal behavioral disruption, securing 10% to 30% productivity gains that Kvarfordt considers meaningful victories. Amplification expands AI across multiple pipeline stages, while Reinvention completely restructures buyer journeys around buyer preference, as exemplified by Jack Dorsey's organizational restructuring across 10,000 employees at Block. Deep reinvention frequently triggers a temporary dip in performance before generating substantial growth, a dynamic Kvarfordt terms the 'slingshot effect.' At Sendoso, restructuring a 15-person SDR team into a lean, AI-centric group caused pipeline creation to dip in month one, match the original baseline by month two, and reach 30% of total company pipeline by month four. Similarly, ZoomInfo replaced inbound intake SDRs with a 1mind autonomous agent, reporting a 16x ROI and outperforming baseline human conversion within five months. Aligning Autonomy Levels and Pilot Kill Criteria (12:46) Deploying agentic AI requires granular process mapping that separates human tasks from data requirements. A workflow audit of standard discovery calls at Momentum revealed 23 manual tasks requiring 3.5 hours across eight systems, generating 180 data points—less than 5% of which sales reps routinely captured. Organizational friction arises when executives demand Level 4 interactive autonomy from teams whose data integration and human workflows only support Level 2 automated drafting. To maintain pilot discipline, Kvarfordt recommends isolating a single CFO-recognized target, mapping the top three human tasks and data points within that workflow, and establishing a strict 30- to 90-day kill date. For enterprise sales with 18-month cycles where closed-won revenue cannot be evaluated quickly, leaders must tie kill dates to upstream leading conversion indicators to decide whether to scale or terminate the pilot. Kvarfordt stresses that agentic AI cannot rescue broken processes, unvalidated customer profiles, or siloed data architectures. While radical reinvention can accelerate operational autonomy, most enterprises achieve durable ROI by optimizing existing workflows, establishing realistic maturity baselines, and remaining willing to terminate underperforming pilots before committing further capital.

Jonathan Kvarfordt（业内常称“Coach K”）指出，企业在落地代理式人工智能（Agentic AI）时屡屡受挫，核心症因并非技术缺陷，而是企业在未明确核心业务瓶颈前便盲目采购工具。他主张企业应摒弃孤立的AI战略，将AI直接锚定于销售速度公式，并依据自身的组织成熟度分级推进落地。 企业AI落地的现实鸿沟与三大失败诱因 (1:47) 2025年，Shopify的全员备忘录、Salesforce宣称30%至50%的工作由AI完成，以及Duolingo冻结可被AI替代的岗位招聘等高层指令，给市场拓展（GTM）管理者带来了巨大交付压力。尽管McKinsey的行业报告声称企业AI采用率高达88%，但Momentum基于数十万次客户实际对话的分析表明，2025年上半年仅有7.6%的企业实现了除ChatGPT或Copilot基础席位之外的真正业务化运营，下半年该比例也仅上升至24%。绝大多数企业只是分发了聊天机器人账号，便仓促标榜完成数字化转型，存在严重的“重开药方而轻诊断”现象。 Jonathan将AI项目失败归结为三大非技术性组织诱因：第一是技术优先思维，管理层过度聚焦于购买哪款软件，却忽视了定位实际操作瓶颈；第二是数据不成熟，缺乏具备安全边界的数据架构，导致销售与营销数据相互割裂；第三是变革管理匮乏，未在前期规划人类员工的适应与行为工作流。例如在Artisan和11x等AI SDR平台的早期实践中，两家同阶段同行业的公司同时部署，一家取得显著成功，另一家完全失败并归咎于软件，但深度诊断显示失败原因实为目标客户画像（ICP）不准确。 销售速度公式与业务结果反向推演 (4:21) Jonathan借鉴在FranklinCovey的经验，强调必须以终为始，构建“运用AI的业务战略”而非独立的“AI战略”。领导者应将AI举措直接挂钩于明确且有时间限制的财务目标（例如在12个月内将营收从2500万美元提升至3500万美元），并通过经典销售速度公式进行反向拆解：销售速度等于商机数量、平均客单价与赢率的乘积，再除以销售周期长度。通过审视具体哪个杠杆滞后，团队能诊断出对应的人力行为与流程缺口，进而针对性地引入AI填补短板。 在组织诊断模型中，问题被划分为四大维度：员工基础技能、整体GTM战略与定位、产品市场匹配度（PMF），以及系统与技术基础。Jonathan强调，如果定位出现偏差或产品缺乏市场需求，AI绝不可能拯救业务，因为技术只是整个业务飞轮中的单一系统环节，无法替代核心战略与产品价值。 OAR落地光谱与成熟度错配陷阱 (7:51) 在实施路径上，GTM AI的落地可划分为优化（Optimize）、放大（Amplify）与重塑（Reinvent）三阶OAR光谱。优化侧重于为既有传统流程提速，例如利用AI SDR加快会议排期，组织变革阻力较低，常适合大型企业；放大则是将AI拓展至销售流程的多个阶段，例如承接7步漏斗中的前3步（需求发掘、初次演示与资质审核）后再转交销售经理；重塑则是彻底以买方体验为中心重构旅程，仅在产生最大价值的节点引入人工销售。 组织在尝试重塑时往往会经历先倒退后跨越的“弹弓效应”。Sendoso在重组15人SDR团队并在首月经历指标下滑后，精简后的AI团队在第四个月贡献了30%的总销售线索；ZoomInfo引入1mind智能体替代入站SDR，在5个月内实现超越基线并斩获16倍ROI。然而，行业中常见的摩擦在于管理层期望以第2级成熟度（内容起草与邮件生成）的组织基础，强行交付第4级（自主买方体验与交互智能体）的结果。企业必须正视当前现实，在未完成底层数据打通与流程清理前，不能盲目拔高自主权期望。 流程拆解、验证止损与长短周期考量 (12:46) Jonathan通过Momentum时期对标准初次销售发现电话的审计展示了流程细化的必要性：一次通话包含23项人工操作，涉及8套独立软件系统和3.5小时的后台行政负担，涵盖12类共180个数据点，而人工销售通常只能记录不足5%的可用数据。为此，领导层在启动试点时应采取务实步骤：确立单一董事会级目标、映射单个工作流中前三项核心人工任务与前三项数据点，并设定30至90天的严格止损日期（Kill Date）。 针对不同销售周期的试点考核，止损标准需动态调整。在长达18个月的大型企业销售周期中，评估不能坐等最终成单，必须在单季度内拆解出驱动长周期的上游先行转化指标进行验证；而在周期仅为5天的快速成交模式下，数据反馈回路几乎是实时的。领导者必须保持自律，避免在缺乏变革准备时被迫盲目激进重塑，因为通过优化旧流程获得10%至30%的效率提升本身就是重大胜利；若在测试期内先行指标未达标，应果断叫停项目，而非持续投入。 总体而言，企业在引入代理式AI时，必须在成熟度基线、数据治理和组织变革速度之间建立理性预期。优化流程往往能更快取得确定性收益，而激进重塑虽能缩短通向高级自主智能的路径，但需要承受巨大的组织重构代价；明确先行指标并坚决执行止损机制，是防范AI试点沦为无底洞投资的核心保障。

## Background 本视频由 Ai4 频道于 2026 年 8 月 19 日发布在 YouTube 平台，主讲人为 1mind 营销副总裁 Jonathan Kvarfordt（Coach K）。视频内容基于 70% 至 85% 的企业 AI 项目因非技术因素在试点期折戟的行业现实，旨在为 GTM 及营销高管提供一套通过 OAR 框架（优化、放大、重塑）明确业务指标、重构买家旅程并实现实际 ROI 的行动指南。 ## TL;DR 演讲指出大部分企业 AI 项目失败并非技术缺陷，而是源于盲目追求工具的技术优先思维、割裂的数据现状以及极度缺失的变革管理。讲者提出“以终为始”的反推方法论，主张将 AI 战略回归到驱动销售周转率（Sales Velocity）等硬性业务指标上，并通过 OAR 模型帮助团队在传统流程优化与彻底以买家为中心的重塑之间找到适合自身成熟度的路径。通过 Sendoso 与 ZoomInfo 等实际案例，演讲强调了人机数据流落差与五级成熟度的现实约束，并建议企业为所有 AI 试点设立基于领先指标的明确止损日期（Kill Date）。 ## Atmosphere & style 整场演讲散发着干练、务实且不留情面的“教练式”诊断风格，演讲者 Coach K 语调沉稳笃定，以第一线操盘者的视角直击行业过度吹捧 AI 普及率的虚妄泡沫。现场氛围从开篇对管理层盲目 FOMO 焦虑的犀利剖析，逐渐过渡到底层数据测绘与系统思维的理性重构，充满了剥除技术幻象后的清醒感与实战压迫感。整场分享节奏紧凑、框架严密，既有直面业务倒退阵痛的坦诚，也给予了业务团队一套能够抵御董事会盲目施压的战略定力。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=nqSb6OeXze8&t=4s) 企业在推进 AI 落地时最常犯的错误是过早纠结于“买哪款工具”，正确的做法是在明确要解决的具体业务问题与战略目标后再评估技术方案。 - [01:47](https://www.youtube.com/watch?v=nqSb6OeXze8&t=107s) 行业报告声称的超高 AI 采用率大多只是采购了席位（如 ChatGPT 或 Copilot），真实业务对话数据显示实际实现业务化运营的比例极低。 - [04:21](https://www.youtube.com/watch?v=nqSb6OeXze8&t=261s) 导致 70% 至 85% 的 AI 试点失败的三大核心根源并非技术本身，而是技术优先思维、数据成熟度不足以及极度糟糕的变革管理。 - [04:21](https://www.youtube.com/watch?v=nqSb6OeXze8&t=261s) AI 落地应坚持“以终为始”，不存在孤立的“AI 战略”，只有以销售周转率等硬核商业指标为牵引的“应用了 AI 的业务战略”。 - [06:52](https://www.youtube.com/watch?v=nqSb6OeXze8&t=412s) 业务失败通常可归因于技能（Skills）、战略（Strategy）、产品契合度（Product）和系统（Systems）四个维度，而 AI 仅属于底层的系统层，无法掩盖战略或产品本身的缺陷。 - [08:24](https://www.youtube.com/watch?v=nqSb6OeXze8&t=504s) OAR 模型将 AI 实施划分为三种策略：优化（Optimize，对现有流程单点提速）、放大（Amplify，接管多阶段任务）与重塑（Reinvent，以买家体验为中心重构旅程）。 - [10:56](https://www.youtube.com/watch?v=nqSb6OeXze8&t=656s) 流程重塑初期往往伴随阵痛，如同拉满弹弓的石子，通过短期的人员重组与流程重构换取长期管线占比与 ROI 的爆发式增长。 - [12:46](https://www.youtube.com/watch?v=nqSb6OeXze8&t=766s) 传统销售首通电话暴露出严重的人机不对称：销售人员需执行 23 项手动操作并耗时 3.5 小时，却只能捕获潜在 180 个数据点中的不足 5%。 - [12:46](https://www.youtube.com/watch?v=nqSb6OeXze8&t=766s) 许多企业感到挫败的根源在于管理层期望获得 Level 4 的自主化智能体产出，但组织自身的数据与工作流成熟度仅停留在 Level 2 水平。 - [15:01](https://www.youtube.com/watch?v=nqSb6OeXze8&t=901s) 组织准备度重于技术本身，在推进试点前必须设定严苛的“止损截止日期”（Kill Date），若 30 至 60 天内领先指标未达预期便果断裁撤。 - [17:45](https://www.youtube.com/watch?v=nqSb6OeXze8&t=1065s) 设定试点评估期限需结合业务模式的客观规律：长销售周期需拆解季度先行指标，短销售周期则可依靠快速闭环迅速调整或终止。 - [20:31](https://www.youtube.com/watch?v=nqSb6OeXze8&t=1231s) AI 绝非万能解药，企业应率先用管理与流程手段诊断并解决低效问题，明确区分是流程优化还是 AI 介入场景，避免盲目把 AI 砸向问题。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=nqSb6OeXze8&t=4s) 错误的问题与演讲者背景介绍 在过去五年的 AI 实践生涯中，演讲者 Jonathan Kvarfordt（人称“Coach K”）被问到最频繁的一个问题始终是：“我到底该买哪款 AI 工具？”这本身并不是一个坏问题，症结在于提问的时机往往完全颠倒。企业管理者总是在还没想清楚要解决什么业务问题时，就急于挑选工具。Jonathan 希望通过本次分享，不仅让大家明白何时才是提出这个问题的正确节点，更为业务团队提供一套下周就能在公司内部落地推行的四步框架。 Jonathan 解释了自己“Coach”（教练）外号的由来：五六年前他在犹他州一家 40 人的初创公司工作，当时团队缺乏多样性，里面竟然有近一半的男性都叫 John。为了消除日常沟通中的混乱，大家纷纷起了绰号，而他当时正担任销售团队的教练，“Coach”便顺理成章地成了他的代称，并一路沿用至今。Jonathan 的履历横跨一线业务与学术界：他曾担任 Momentum 的战略与营销副总裁（该公司已被 Salesforce 收购，但他因热爱初创氛围而选择加入 1mind）；于 2022 年创立了 GTM AI Academy；目前担任 1mind 的营销副总裁；同时还在布莱恩特大学（Bryant University）面向职场专业人士讲授 AI 进阶课程，并为 Spiky、Luster、Recap、GTM Buddy 等多家企业提供咨询。 ```mermaid flowchart LR accTitle: Jonathan Kvarfordt 职业背景与经验沉淀 subgraph S1["行业落地经验"] A["Momentum 战略与营销 VP<br/>(后被 Salesforce 收购)"] B["1mind 营销 VP<br/>(智能体 Agentic AI 落地)"] end subgraph S2["教育与生态布道"] C["GTM AI Academy 创始人<br/>(2022年创立)"] D["Bryant 大学特聘导师<br/>(AI 专业实战课程)"] end S1 --> E["企业级 AI 落地实践框架"] S2 --> E ``` --- ### [01:47](https://www.youtube.com/watch?v=nqSb6OeXze8&t=107s) 行业狂热与 AI 落地的残酷现实差距 从 2025 年开始，企业董事会自上而下推动 AI 的压力达到了前所未有的峰值。先是 Shopify 在 4 月发布了一份震动业界的内部备忘录，逼得所有公司惊呼“天哪，我们必须做点什么”；紧接着 Salesforce 首席执行官 Marc Benioff 宣称公司内部已有 30% 到 50% 的工作交由 AI 处理；随后 Duolingo 也发出通告，明确表态“除非证明某项工作 AI 无法胜任，否则一律冻结招聘”。面对铺天盖地的头条新闻和董事会下达的 ROI 军令状，绝大多数 Go-To-Market（GTM）管理者都陷入了“唯恐落后于人”的极度焦虑中。但 Jonathan 明确指出：“大家其实都处于落后和摸索阶段，没有谁真正领先。” 麦肯锡等咨询机构的研究曾声称企业 AI 采用率已高达 88%，但 Momentum 对数十万场企业客户对话的实际分析却揭示了残酷的真相。在现实业务中，绝大多数团队不过是采购了 ChatGPT 或微软 Copilot 账号，就自欺欺人地将其标榜为“业务转型”。整个行业充斥着“开药方的人太多，真正做诊断的人太少”（"Too many prescriptions and not enough diagnosis"）的怪相。 | 统计口径 | 2025 年上半年数据 | 2025 年下半年数据 | 核心实质 | |:--- |:--- |:--- |:--- | | **行业咨询报告宣称** | 约 88% 的采用率 | 持续高位 | 将购买软件席位（如 ChatGPT/Copilot 许可）等同于 AI 落地 | | **真实业务对话分析** | 仅 7.6% 真正实现业务化运营 | 增长至 24% | 多数企业尚未建立适配 AI 的底层架构与工作流 | --- ### [04:21](https://www.youtube.com/watch?v=nqSb6OeXze8&t=261s) AI 项目失败的三大根源与“以终为始”的方法论 大部分 AI 项目之所以无法走出试验期，核心原因并非技术缺陷，而是组织管理问题。具体表现在三个维度： 1. **技术优先的盲目思维（Tech-first mindset）**：一上来就纠结买哪家技术； 2. **数据成熟度极低（Data immaturity）**：缺乏安全边界与防护栏架构，各类业务系统（营销、销售对话等）彼此割裂成为数据孤岛； 3. **极度糟糕的变革管理（Atrocious change management）**：完全没有规划员工行为与业务流程的迁移方案。 Jonathan 举了一个生动的对比案例：当 Artisan 和 11x 等 AI SDR（AI 销售开发代表）产品风靡时，两家同行业、同体量的公司同时引入了该技术。一家运转良好，另一家却彻底崩盘并甩锅给 AI。深入诊断后发现，失败公司的真正根源在于自身的目标客户画像（ICP）设定严重偏差，AI 只是加速放大了既有的战略失误。 借鉴富兰克林柯维（FranklinCovey）的经典法则，AI 转型必须坚持“以终为始”（"Begin with the end in mind"），从战略业务结果反向倒推。以提升营收和管线（Pipeline）为例，管理层应聚焦经典的“销售周转率”（Sales Velocity）模型，通过拆解四个核心变量来定位增长杠杆： $$\text{销售周转率 (Sales Velocity)} = \frac{\text{商机数量 (Number of Ops)} \times \text{平均客单价 (ACV)} \times \text{赢单率 (Win Ratio)}}{\text{销售周期天数 (Sales Cycle)}}$$ ```mermaid flowchart LR accTitle: 战略驱动的 AI 实施反推链路 A["战略业务目标<br/>(如：营收/管线增长)"] --> B["核心运营指标<br/>(销售周转率四大变量)"] B --> C["底层业务流程与行为<br/>(任务拆解与行为分析)"] C --> D["差距分析<br/>(定位业务断点)"] D --> E["定制 AI 落地策略<br/>(解决具体瓶颈)"] ``` --- ### [06:52](https://www.youtube.com/watch?v=nqSb6OeXze8&t=412s) 业务战略先行与四维归因框架 设定目标时必须满足三个硬性标准：可量化（Quantifiable）、有明确截止日期（Time-bound）以及具备实质业务影响力（Impactful）。例如设定“在 12 个月内将年营收从 2500 万美元提升至 3500 万美元”。团队需要基于当前销售周转率测算达到新增 1000 万美元所需的指标跃迁，并找出具体差距，再思考 AI 如何弥补缺口。Jonathan 强调：“根本不应该存在所谓独立的‘AI 战略’，只存在‘应用了 AI 的业务战略’。” 当技术无法转化为实际业务成效时，问题通常出在四个维度，而 AI 仅仅隶属于最末端的系统层： - **人员技能（Skills）**：团队是否具备执行新业务流程的专业能力； - **整体战略（Strategy）**：目标客户画像（ICP）与市场定位是否精准； - **产品契合度（Product / PMF）**：产品是否真正满足市场需求； - **支撑系统（Systems）**：底层工具链与技术架构（AI 仅属于此范畴）。 如果企业连基本的 ICP 都找不准，或者产品本身缺乏 PMF，盲目引入 AI 工具不仅救不了公司，反而会导致资源的大量浪费。 --- ### [08:24](https://www.youtube.com/watch?v=nqSb6OeXze8&t=504s) OAR 模型：从优化现有流程到重塑买家体验 面对 AI 变革，企业通常存在两种截然不同的思考路径：一种是沿着传统销售流程修修补补，试图做得更快、更便宜；另一种则是运用系统思维（Systems Thinking）彻底颠覆传统链路。Jonathan 为此提出了 **OAR 模型**（Optimize, Amplify, Reinvent），企业应根据自身组织成熟度与风险承受力来匹配对应策略： - **Optimize（优化现有流程）**：沿用老旧打法，仅用 AI 提升单点速度。例如“用 AI 自动发邮件约会议”，本质仍是传统销售思维。 - **Amplify（放大关键环节）**：让 AI 接管多阶段销售动作。例如在典型的七步销售流程中，让 AI 自主承担需求挖掘（Discovery）和产品演示（Demo）等前三步，随后再无缝移交给人类销售。 - **Reinvent（重塑买家旅程）**：从“以销售为中心”彻底转向“以买家为中心”，摆脱人类销售产能上限的束缚，在买家偏好与机器交互的场景全面引入 Agentic AI，让人类在最能发挥价值的高光时刻介入。 ```mermaid flowchart TD accTitle: OAR 框架光谱演进 subgraph O["Optimize (优化)"] O1["沿用传统销售流程"] O2["AI 替代单点重复劳动<br/>(如：AI 自动约会)"] end subgraph A["Amplify (放大)"] A1["多阶段任务扩展"] A2["AI 自主完成需求调研与演示<br/>(覆盖 7 步销售法的前 3 步)"] end subgraph R["Reinvent (重塑)"] R1["以买家体验为核心构建"] R2["Agentic AI 全流程自主响应<br/>人类销售仅在关键决策点介入"] end O --> A A --> R ``` --- ### [10:56](https://www.youtube.com/watch?v=nqSb6OeXze8&t=656s) 转型阵痛与实际落地案例：Sendoso 与 ZoomInfo 在业务流程重塑的初始阶段，团队往往会产生“业务变慢了甚至倒退了”的错觉。Jonathan 将这种现象形象地比喻为“拉满弹弓的石子”——后撤是为了获得更惊人的爆发力： - **Sendoso 案例**：原先由 15 名 SDR 贡献团队 15% 的销售管线。管理层决定打破现状，第 1 个月对团队实施重组，表现不佳者淘汰或转岗为 AE，抽调对 AI 有极高热情的人员从底层重构工作流。第 1 个月业务指标看似下滑，但到了第 2 个月，精简后的极少数 BDR 借助 AI 实现了与过去 15 人相同的产出；到了第 4 个月，SDR 贡献的管线占比飙升至 30%，营销整体节奏实现了超速爆发。 - **ZoomInfo 案例**：原本部署了大量负责处理官网集客线索（Inbound SDR）的人员。公司将其转型调整至战略大客户开发或出海拓客（Outbound），并在官网原有位置部署了 1mind 的“超级数字人”（Superhuman）。在完全基于买家需求重塑交互后，仅用 5 个月时间便斩获了高达 16 倍的 ROI。 --- ### [12:46](https://www.youtube.com/watch?v=nqSb6OeXze8&t=766s) 销售第一通电话的人机数据差距与五级成熟度模型 为了证明传统流程的低效与数据断层，Jonathan 分享了他在 Momentum 期间对销售首通电话（First Call）的详尽测绘数据。在人类销售与数据流两个侧面，存在着极其惊人的不对称： ```mermaid flowchart LR accTitle: 销售首通电话的人机负担对比 subgraph Human["人类销售负担"] H1["23 项手动操作"] H2["耗时 3.5 小时"] H3["跨越 8 个独立系统"] end subgraph Data["数据资产流失"] D1["包含 180 个潜在数据点"] D2["横跨 12 个业务类别"] D3["人类销售仅能捕获不足 5%"] end ``` 在 AI 的落地进阶上，Jonathan 将企业划分为五个层级（Level 1 至 Level 5）： - **Level 1**：基础建议（如 AI 文本修正与提示）； - **Level 2**：辅助起草（如自动生成邮件草稿与会议纪要）； - **Level 3**：智能体工作流执行（Agentic Workflows，AI 具备一定的多步骤协同能力）； - **Level 4**：高度自主化的买家交互体验（面向买家端自主闭环）； - **Level 5**：跨业务系统的全自主协同生态。 许多企业之所以挫败，核心矛盾在于“管理层期望获得 Level 4 的自动化产出，但组织自身的数据和流程成熟度仅仅停留在 Level 2”。 --- ### [15:01](https://www.youtube.com/watch?v=nqSb6OeXze8&t=901s) 准备度原则与本周即可落地的三项行动指南 “组织的准备度永远重于技术本身”（"Preparedness trumps technology"）。在正式上线任何 AI Pilot（试点项目）之前，企业必须建立明确的“止损/终止标准”（Kill Criteria）。虽然 30 天内可能无法直接看到最终收入的变化，但在 30 至 60 天内必然能监测到某些领先指标（Leading Conversion Metrics）的迁移；一旦在预设周期内无法达到指标，就必须果断叫停该试点。 Jonathan 为业务团队总结了本周即可启动的三个具体行动步骤： - **锁定一个 CFO 认可的硬核商业指标**：无论是营销、RevOps 还是销售赋能团队，明确一个能直接体现商业价值的目标； - **梳理单条业务工作流并做极简萃取**：不要被 180 个数据点吓倒，仅挑出最重要的“3 个核心数据点”和“3 项人类核心任务”进行标准化改造； - **为所有进行中的 AI 试点设定明确的截止日期（Kill Date）**：设定 30 天或 90 天的严苛评估窗口，以领先指标为准绳，达不到预期便果断止损裁撤。 ### [16:51](https://www.youtube.com/watch?v=nqSb6OeXze8&t=1011s) 根据组织现状选择策略：从渐进式优化到彻底重塑 并非所有企业都必须立刻进入“彻底重塑”（Reinvent）的阶段。如果组织或团队尚未做好准备，盲目追求颠覆性变革往往适得其反。能够借助 AI 实现 10% 到 30% 的效率提升本身就是一项巨大的胜利。企业应当坦然接受自身所处的阶段：若现阶段的能力仅能支持“优化”（Optimize），那就先专注把优化做好，打稳基础后再逐步向上推进。 ```mermaid flowchart TD accTitle: OAR Framework Progression subgraph OAR["OAR 实施路径与组织成熟度"] O["优化 (Optimize)<br/>现有流程提速 10%-30%"] --> A["放大 (Amplify)<br/>扩展团队产能与触达"] A --> R["重塑 (Reinvent)<br/>重构组织架构与业务流"] end subgraph Foundation["底层支撑要素"] D["数据编排 (Data)"] H["人员协同 (Humans)"] S["系统架构 (Systems)"] end Foundation -.-> OAR ``` 与此同时，对于那些决心采取激进举措的组织，行业中也存在截然不同的范式。例如 Jack Dorsey 领导下的 Block（前身为 Square），在其 10,000 人的公司规模下裁员 4,000 人并全面重构组织层级，通过打破旧体制来实现颠覆式重塑。选择何种路线完全取决于组织自身的特点和团队的承载力。关键在于对所选的层级保持笃定，“不必为没有立刻开启转型而焦虑，因为在没有准备好的情况下盲目转型，从长远来看没有任何帮助”。 - **务实的生产力预期**：在成熟度不足时，将目标定在 10%~30% 的效率提升同样属于实质性成功。 - **激进重组的行业案例**：Block 公司通过大规模组织瘦身与层级重构，展示了大型企业彻底打破旧体系（"blowing up the entire org"）的重塑路径。 --- ### [17:45](https://www.youtube.com/watch?v=nqSb6OeXze8&t=1065s) 问答环节：如何设定 AI 项目的“终止日期” 在问答环节中，现场观众针对演讲中提及的“项目终止日期”（Kill Date）提出疑问，询问该如何具体设定这一时间节点以及有哪些决定性考量因素。对此，设定期限的核心依据在于企业追求的业务结果以及业务周期的客观规律。 不同商业模式的反馈闭环周期存在巨大差异。对于大型企业级服务（Enterprise Motion）而言，销售周期往往长达一年半，此时不可能在单季度内直接看到最终成单数字的变化，团队必须拆解出能够在季度（Quarter）内有效衡量的先行指标（Leading Metrics），以此评估 AI 工具是否真正对长达一年半的销售周期产生了积极拉动。相反，对于销售周期仅有 5 天的客户，反馈循环极快，团队可以迅速获得成效验证并据此调整或终止项目。 - **长销售周期（约 1.5 年）**：依赖季度级中间指标来衡量工具对全流程推进的拉动作用。 - **短销售周期（约 5 天）**：依靠极短的数据反馈闭环迅速验证假设、调整流程。 --- ### [18:17](https://www.youtube.com/watch?v=nqSb6OeXze8&t=1097s) 问答环节：AI 成熟度等级与董事会期望管理 现场观众进一步提问：如果企业向董事会承诺达到 Level 4 的自主化智能体（Autonomous Agent）成熟度，但实际落地时由于数据或流程缺陷，性能表现仅停留在 Level 2 水平，该如何向董事会解释这种成熟度与组织就绪度之间的落差，以及人类与 AI 之间的责任归属（Liability）问题。 针对成熟度层级落差的问题，任何 AI 能力的发挥都受限于企业现有的系统架构、底层数据编排以及在此期间人类员工的协同方式。这些基础工程与管理变革绝不可能在短短一个月内凭空完成。如果管理层在立项 30 天后就质问为什么转型只达到 Level 1 或 Level 2，团队必须向董事会清晰说明：实施路径取决于选择的是优化（O）、放大（A）还是重塑（R）。彻底重构流程（"blow things up"）的企业或许能更快迈向 Level 4，而渐进式优化的企业则需要更长的爬坡期。团队需要帮助董事会理解，启动 AI 转型不仅是引入模型，更需要时间理顺人员、数据与流程的配合，而不是盲目“把 AI 砸向问题”。由于演讲时间所剩无几，针对深层的 AI 与人类责任界定问题，讲者建议在会后展开一对一深度探讨。 - **落地周期的客观规律**：系统对接、数据流打通与人机协作机制的建立需要周期，难以在 30 天内一蹴而就。 - **路径对成熟度速度的影响**：彻底颠覆旧流程的重塑路线往往比在旧框架内的优化路线能更快达到高阶自主化。 --- ### [20:31](https://www.youtube.com/watch?v=nqSb6OeXze8&t=1231s) 问答环节：业务流程优化、数据演进与落地时间表 最后一位观众提出，在执行“优化”（Optimize）策略并设定截止日期时，企业是否往往需要先梳理并改进既有流程本身，甚至在未引入 AI 之前就需要做流程优化；如果业务流程和数据指标处于动态变化中，设定的时间线是否会受到影响。 针对这一问题，核心原则在于明确“AI 并非万能解药”。在很多场景下，流程本身的低效与 AI 无关，需要率先通过纯管理或业务手段进行优化；这也是为什么必须“从业务结果出发”（Outcomes First）来界定问题——首先判断这究竟是不是一个需要 AI 介入解决的问题。在定义上，“优化”意味着在不改变既有流程框架的前提下让流程跑得更快，而“重塑”则是打破并重建流程。无论是优化还是重塑，流程与数据的调整都需要耗费时间，设定时间框架的本质是为组织划定一个观察窗口：明确我们需要在多长时间内维持这套流程运行，以充分验证它是否正在朝着预期的业务结果迈进。 - **以目标驱动选型**：先确认业务瓶颈的本质，避免把纯流程问题误当成 AI 问题来处理。 - **优化的本质定义**：优化是使现有流程加速运转，而非直接推倒重来。 - **时间周期的真正作用**：时间框架为流程验证设定了边界，用于衡量方案在特定周期内是否真正交付了预期的商业结果。 ## Notable quotes > [00:04](https://www.youtube.com/watch?v=nqSb6OeXze8&t=4s) 我到底该买哪款 AI 工具？ — Jonathan Kvarfordt > [01:47](https://www.youtube.com/watch?v=nqSb6OeXze8&t=107s) 大家其实都处于落后和摸索阶段，没有谁真正领先。 — Jonathan Kvarfordt > [01:47](https://www.youtube.com/watch?v=nqSb6OeXze8&t=107s) Too many prescriptions and not enough diagnosis — Jonathan Kvarfordt > [04:21](https://www.youtube.com/watch?v=nqSb6OeXze8&t=261s) Begin with the end in mind — Jonathan Kvarfordt > [06:52](https://www.youtube.com/watch?v=nqSb6OeXze8&t=412s) 根本不应该存在所谓独立的‘AI 战略’，只存在‘应用了 AI 的业务战略’。 — Jonathan Kvarfordt > [15:01](https://www.youtube.com/watch?v=nqSb6OeXze8&t=901s) Preparedness trumps technology — Jonathan Kvarfordt > [16:51](https://www.youtube.com/watch?v=nqSb6OeXze8&t=1011s) 不必为没有立刻开启转型而焦虑，因为在没有准备好的情况下盲目转型，从长远来看没有任何帮助 — Jonathan Kvarfordt > [20:31](https://www.youtube.com/watch?v=nqSb6OeXze8&t=1231s) AI 并非万能解药 — Jonathan Kvarfordt ## Mentioned Jonathan Kvarfordt, 1mind, Shopify, Salesforce, Marc Benioff, Duolingo, McKinsey, Momentum, ChatGPT, Microsoft Copilot, Artisan, 11x, FranklinCovey, Bryant University, GTM AI Academy, Spiky, Luster, Recap, GTM Buddy, Sendoso, ZoomInfo, Block, Square, Jack Dorsey ## Video description Jonathan Kvarfordt - VP, Marketing, 1mind Most go-to-market teams buy agentic AI before they have defined what it is supposed to fix. Industry research puts the failure rate of AI initiatives as high as 70 to 85 percent, and the reasons are almost never technical. This session hands revenue and marketing leaders the foundation that separates the projects that pay off from the ones that quietly die in pilot: a business target your board will fund, an honest map of where the buyer journey leaks revenue today, and a method for matching agentic capability to the moments that actually move the number. You will leave with the OAR framework used to take a GTM org from "we bought some tools" to measurable return, drawn from the same playbook behind 100 Superhumans running live across 70+ brands. What you will leave with: 1) A way to set an agentic AI target your board will actually fund, tied to revenue, cost, or risk, with a deadline attached. 2) The OAR model for sorting use cases into Optimize, Amplify, and Reinvent, so you start where the payoff comes fastest. 3) A pilot-to-scale sequence that dodges the three reasons initiatives fail: tech-first thinking, messy data, and no change management. 4) A candid read on the metrics that shift once agentic AI works, including why fewer meetings can mean more pipeline. [Source](https://www.youtube.com/watch?v=nqSb6OeXze8)

[0:04] Jonathan Kvarfordt: 我到底该买哪款 AI 工具？ [1:47] Jonathan Kvarfordt: 大家其实都处于落后和摸索阶段，没有谁真正领先。 [1:47] Jonathan Kvarfordt: Too many prescriptions and not enough diagnosis [4:21] Jonathan Kvarfordt: Begin with the end in mind [6:52] Jonathan Kvarfordt: 根本不应该存在所谓独立的‘AI 战略’，只存在‘应用了 AI 的业务战略’。 [15:01] Jonathan Kvarfordt: Preparedness trumps technology [16:51] Jonathan Kvarfordt: 不必为没有立刻开启转型而焦虑，因为在没有准备好的情况下盲目转型，从长远来看没有任何帮助 [20:31] Jonathan Kvarfordt: AI 并非万能解药

[0:04] Jonathan Kvarfordt: What AI tool should I buy? [1:47] Jonathan Kvarfordt: too many prescriptions and not enough diagnosis. [4:21] Jonathan Kvarfordt: Organizations should never build an isolated "AI strategy," but rather a "business strategy that uses AI." [7:51] Jonathan Kvarfordt: Undergoing this reinvention often creates a temporary sensation of moving backward before dramatic growth occurs—a dynamic Jonathan describes as the "slingshot effect." [16:51] Jonathan Kvarfordt: Securing a 10% to 30% productivity increase through optimization alone represents a substantial and legitimate victory. [17:45] Jonathan Kvarfordt: In an enterprise motion with an 18-month sales cycle, establishing an evaluation window cannot rely on closed-won revenue alone. [18:17] Jonathan Kvarfordt: Autonomous performance is heavily constrained by existing tech stacks, data orchestration quality, and the defined role of human operators. [20:31] Jonathan Kvarfordt: Process enhancement often reveals that "AI is not the answer" because the underlying bottleneck has nothing to do with technology.

[0:04] This question is the one I get asked more often  than any other one in the last five years of doing [0:09] AI, which is my nickname is coach, and I'll tell  you why in a second, but I have people ask me, [0:14] what AI tool should I buy? And this is not a  bad question. I just usually get asked this at [0:21] the wrong time. So, what I want to do today, if  my slides work with me, is I want to make sure [0:28] that you will leave knowing when the right time  is to ask this question and then also give you [0:33] four things or a framework you can use hopefully  this next week you can apply inside of your own [0:38] company. Um, my name is Jonathan Carford. My  nickname is Coach K. Short story, I was in a [0:44] startup five, six years ago and it was in Utah,  which means it's not very diverse. And half of [0:50] us were there's 40 guys or 40 girls and guys. Half  of us were named John. And so to reduce confusion, [0:57] we all got nicknames. And mine was coach because I  was a sales coach of the team. So it just kind of [1:00] stuck because everywhere I've gone since someone  known me as coach. So my nickname is coach. And [1:05] I'm currently a VP of marketing at OneMind. I  also founded the GTMai academy in 2022. Um I [1:11] was the VP of strategy and marketing at Momentum,  which was recently acquired by Salesforce four or [1:16] five months ago or so. And um I had the option  of going to Salesforce but I love the startup [1:21] world. So I chose to come back to startup with one  mind and the team there which is amazing. And then [1:26] I work with um I work with all these different  companies like uh Spiky and Luster and Recap and [1:30] GTMB buddy and others as well as I teach at Bryant  University. They're their professional level AI [1:36] classes for professionals. So I live, breathe,  eat, sleep, teach, talk AI all the time and I love [1:41] it and can't get enough. So really excited to be  here and I appreciate you all being here with me. [1:47] All right. So, I want to kind of go back and look  at more holistically where we've come from. And [1:51] that kind of leads us to where we are. So, in in  2025, we saw this push when I was at Momentum. [1:57] We saw this change in 25 that made this board  level push for AI more than ever. And it started [2:02] in April. Does anyone remember this memo? Shopify  came out and everyone's like, "Holy Hannah, we got [2:09] to do something." So this was what kind of started  this trend of push from upper levels down to say [2:14] figure out AI in our business. The next part was  when you had things like Mr. Ben off here saying [2:20] that his company was doing 30 to 50% of the work  with Salesforce or with AI inside of Salesforce [2:26] which is crazy. Then you have another memo from  good old Dualingo where he talks about we're going [2:32] to focus on not hiring anybody unless we can prove  that AI can't do it before. So we have all this [2:37] pressure. So we as go to market leaders see the  headlines. We get the pressure from our leaders [2:41] and boards of saying figure out AI and get ROI  because everyone else is doing it. So you got to [2:45] figure it out too. So anyone else feel like I'm  behind and guess what? So is everybody else. I [2:53] can tell you right now everyone's behind. So if  you feel like this, please feel comforted that [2:57] like we're all figuring this out and pioneering  new ways. And I'm hopefully take you through to [3:02] help you get ahead a little bit. When I was at  Momentum, we did this analysis on conversations [3:09] of people we were talking to because all we talked  to was small business to enterprise companies, [3:13] revops and AI ops on how to get AI inside the  company. And so we did this measurement and I [3:18] measured it against the McKenzie study that talked  about at the time last year we are 88% adoption of [3:24] AI. Now when we looked at our analysis of hundreds  of thousands of conversations in the last of 2025 [3:32] the pattern I found was at the beginning of  the year the reality was 7.6 were actually [3:37] operationalized or could be used meaning that they  actually had AI beyond just a chatbt license and [3:43] in H2 or half second half of the year it went to  24%. So still going up but it was not as much as [3:48] what Mackenzie was claiming. And what I found was  as we talked to teams, they had mostly co-pilot [3:54] chatbuchi licenses and that was what they were  considering transformation. Does this ring a [3:59] bell for anyone at all? It does for me of people I  talked to. Most revenue and go to market teams are [4:06] not ready or not built for the current AI world.  Mostly because we're all trying to figure this [4:10] out. So I feel like there's too many prescriptions  and not enough diagnosis is the core part of this. [4:21] The reason why I think most AI initiatives fail  comes in three specific places. One is in the [4:27] uh tech first mindset. What AI tech should I buy?  That's number one. Two is data immaturity. They [4:34] don't have a data um architecture that's  both secure and has guardrails for AI to [4:39] access or not. And then you don't have data  at all around different topics from sales [4:43] conversations to marketing to whatever. And  those data points don't talk to each other. [4:47] And then lastly is atrocious change management.  What no matter what you do in AI, I see a lot of [4:53] times where people don't plan ahead for change  management enough so that the adoption and the [4:56] impact doesn't happen. Usually these problems are  not technical. They're organizational issues. So [5:03] as I remember when AI STRs came out and we had  Artisan 11X, everyone's going crazy. Um I had a [5:08] lot of people who got these technologies because  it was the hot thing to get. And the problem was [5:12] when I had when I consulted a specific company had  one of these uh technologies. When we dug into it, [5:18] I had two companies side by side. One was using an  AISDR working really well and the other one about [5:23] the same industry, same size was bombing crazy and  they blamed the AI. When I looked at the analysis, [5:29] it was an ICP target problem. It was not the AI,  but they were blaming it on the AI because they [5:34] weren't ready, were immature to actually have  a technology. I'm going to take you through [5:38] something I've done many times with a lot of  companies where I used to work at Franklin CVY and [5:43] we're obsessed with starting with the end in mind.  So I believe it starts with focusing on strategic [5:47] outcomes first, not the problem outcomes. What  are we trying to go for? Just for this example, [5:52] let's use the example of just overall pipeline  of revenue. Then let's take a step backwards and [5:57] look at metrics. A classic one to look at is sales  velocity. It's one of my favorites out there. Has [6:01] these four metrics. Number of ops, ACV, uh sales  cycle, and win ratio. Those four put together tell [6:08] you how efficient either a person, source or um  avenue or industry is performing inside of your [6:13] revenue. And once you can understand which one of  these can be moved, it drastically impacts your [6:18] revenue or pipeline goals. So once you know where  are we at now and where do we want to be, then you [6:24] look backwards and say, okay, what's the process?  What's the task? What's the behaviors that we're [6:29] looking for that will impact or drive these  metrics that will drive the revenue? And then [6:34] you do a gap analysis to figure out where are we  like what's the problem and how is this impacting [6:39] and that informs your AI strategy. This is how  you start to get ROI is being able to attach [6:44] AI programs directly to strategic outcomes instead  of just guessing and hoping for something to work. [6:52] So this is an example I'll show you real  quick of a real target. I'll send you the [6:56] slides. I promise you you'll take them. Yeah.  Um so your target should be something tied to [7:01] business value. So you have quantifiable, time  bound and impactful. And again, it's going to be [7:07] uh decided by each company. Could be churn, could  be pipeline, could be whatever you want to decide [7:11] on. And then for examples here, it's like saying,  okay, we want to go from revenue of 25 million to [7:16] 35 million in 12 months. That's our goal. So then  you work backwards from revenue saying, okay, [7:21] to add $10 million, what's our sales velocity at  now? And what does it have to be to hit that goal? [7:26] And where's that gap? And how can we help mitigate  that gap? The most successful AI transformations [7:33] don't start with tools. They start with targets.  I firmly believe, even though I'm an AI nerd, [7:38] that there is no such thing or shouldn't be a  such thing as an AI strategy. It needs to be [7:41] a business strategy that uses AI. So, here's the  the problems I see happen um inside of AI. There's [7:51] four main places I see with with trying to get  a technology to impact. It falls under usually [7:56] skills of a human, overall strategy, product  meaning product market fit, and then systems. [8:02] Only one of these has to do with AI. Obviously,  AI can help with all of the rest of them, but one [8:07] of them is the core. And as you can see, I don't  know if you can see the bottom of this, but like [8:11] um if your ICP is wrong, if your targets wrong, AI  is not going to save you. If your product doesn't [8:18] have a right fit and you're convincing yourself  that it does, AI is not going to save you. [8:24] So there's two realities I see over and over with  AI which is some companies look at the old way of [8:30] doing things. The old sales process, the old  buyer journey and they say how can I optimize [8:34] and make this faster, better, cheaper, etc. And  the companies I see making drastic impacts are [8:39] the ones who are thinking differently. How do I  transform this in a totally different way with [8:44] jobs we've done in systems thinking? And there's  I have a spectrum I like to call the Our model. [8:50] So depending on the company um you'll this kind  of has to do with how much risk you have which [8:55] I'll talk about in a second but there's optimize  amplify and reinvent depending on the maturity of [9:01] your org and how large it is and what you're going  through. You may start at the O and you may start [9:06] at the R. Kind of depends on where you're at.  I put it on a spectrum like the less change in [9:11] behavior in systems will put you on the left side  which is usually enterprises because it takes time [9:16] and the more change in behavior in systems usually  is startups is on the right side. So depending on [9:21] how drastic the changes will make impact on it and  you can't see this because of my transcription but [9:27] on the left side is the am I doing this right?  Right side is the O A is in the middle R is on [9:34] the side. So it all depends on again your comfort  level of how much you want to transform it up. [9:38] An example to give you is optimize. You have an  AI that helps you book meetings. That's the old [9:43] way of saying we need meetings. We have an AI do  it for us. It's the old way just make it faster. [9:48] Then there's amplify which is thinking but what  if the AI could do discovery and what if AI could [9:53] do demos and what if AI could go so if you have  a sevenstep sales process what if AI could cover [9:59] three of those not just booking meetings and then  it gets handed off to a human later on the sales [10:04] process. And then lastly is reinvent where you  rethink the entire journey of the buyer for the [10:10] benefit of them will help you be able to win and  then you optimize for what you're supposed to be [10:14] doing hopefully in the first place which is  helping your buyers be successful. So the the [10:20] two sides of this again like one is doing things  the old way and or the new way. The other part [10:25] of it is are you optimizing for your seller  which again this is not a bad thing. This is [10:28] all I did at momentum. Are you optimizing for  the seller or are you optimizing for the buyer [10:34] and releasing yourself from the pressure of of  the of the ceiling of what a seller can do? Um, [10:39] I'm not in the business of saying you should  replace humans. What I am in the business of [10:42] is saying what is the buyer experience and how  can we make that better so that humans come into [10:46] the process at the best point and the AI can  help throughout the other parts where buyers [10:51] would prefer an AI versus a human. Um, okay. So,  I have a lovely picture of this lovely woman. I [10:58] have no idea what she is. Um, but it feels like  when you go through these processes that when you [11:03] start to do this reinvention or AI strategy, you  feel like you're going backwards or you're slowing [11:07] down. The truth is you're this rock that about  to be slingshotted off really quickly because [11:11] every team that I see do this that takes the time  and slows down accelerates way faster afterwards. [11:18] An example is um, uh, Senoso. So the friends  of mine, they came on to my podcast and they [11:23] talked about what they did. They started off their  entire pipeline of 100%. 15 of it was coming from [11:29] STRs when there's 15 of them and they said we  got to do things differently. So month one they [11:35] blew it up. They either moved people out cuz they  weren't performing well or they moved them into a [11:39] um AE position and then they took the few people  who are really geeking out in AI and said we got [11:45] to figure this out from the ground up. So the  first month they took at all of their processes [11:49] and figured out what is the best journey to go on  and how do we optimize that and by month two BDRs [11:56] were doing the same amount as 15 at month two but  during the first two months it felt like they're [12:01] going backwards. So if you looked at your metric  at month one you're like this is not going well [12:06] at all but at month two they hit parody at the  same time. By month four SDRs were doing 30% of [12:13] pipeline with a marketing team that had not slowed  down. So it had accelerated which is awesome. Next [12:18] version I'm use one from one mind is zoom info.  So they had a team of inbound uh strers who were [12:24] doing things on on the website and taking leads  in and what they did is they transition those [12:28] people to either strategic accounts or to outbound  and put a one mind superhuman in its place and in [12:33] five months it was outperforming and getting a  16x ROI in five months by just thinking about [12:38] what would the buyer want to have in this place  versus what we're doing now which is fascinating. [12:46] So there's there's two sides. I'm going to  show you a couple of diagrams which I find [12:49] interesting. There's the human side and there's  the data. And as you start looking through what [12:53] you're going through, you have to always ask,  what are we asking humans to do and what are we [12:56] asking data to do? So I'm going to use an example  of a first call, first or second call of a sales [13:02] process. I went through and I mapped this out  for momentum because we had an automation tool [13:06] that worked with sales teams on first calls. And  I mapped out every single process and data point [13:11] or system that someone uses on average based on  our customers. This is a visual of what that looks [13:16] like. 23 manual tasks, three and a half hours,  eight systems for the human on the data side, [13:23] which you can't see this. It says 180 data points,  12 categories. Most teams get maybe 5% of this on [13:32] one call. We're not even talking on an enterprise  motion. We have multiple calls over six or seven [13:37] months. like this is just one call of data that  we're missing. So, it's thinking about what is [13:42] the most important amount of this. I'm not saying  you should go out and get all of this right now, [13:45] but it's really looking at what is the data we're  really looking for. What's the most important [13:50] things the humans need to focus on that will make  impact on revenue and and really focus on that. I [13:57] want to talk about this real fast because I think  it's interesting. When I talk to um companies that [14:01] I put them into a scale of maturity, I guess  you could say there's level one through five. [14:06] One is when you have like basic suggestions  from AI. Two is when you have like it starts [14:11] to do drafting and emails and that kind of stuff.  Three is when you get to more um agentic workflows [14:16] and execution. But the difference between the  two sides is that the one side's optimizing for [14:20] the human. The other side starts optimizing  for the AI and the buyer. And at level four, [14:25] you start to have this autonomous um experience  with AI and level five is when you have full [14:30] autonomy across different systems. The challenge  I see is that companies are wanting or expecting [14:35] a level four output, but they're only ready for  level two performance. And so the misaligned [14:40] expectations make it people frustrated  because they're looking for a level four, [14:43] but they're really not even ready for it. So  just so you all know, if you ever have to um [14:48] do any type of AI technology implementation, it's  really being honest with your leaders and saying, [14:54] "We're actually level two. You can't expect  this yet." So you got to build up to that. [15:01] Okay. Oh, this works for me. Okay. I believe the  preparedness trumps technology. And too many times [15:07] the whole key of this is that we're just not  prepared enough. We're not ready for it. So, [15:12] in this process, there's there's very quickly I  take you through a a demo process. But before you [15:17] do pilots, if you walk through this, um, if your  decision was we need to scale the capacity of our [15:23] account executives and so you figure out, okay,  we need to figure out how do we get our IP into [15:27] the orchestrate the right ICP in the right places.  And then we have this autonomy level. So we start [15:32] at level two and we start moving up. You want to  be able to match what's happening. So you have a [15:37] kill criteria at a certain point with pilots. So  you know, okay, do we have a metric? Some pilots, [15:42] you're not going to know if it's going to impact  revenue in 30 days, but you will see a conversion [15:47] shift something in 30 to 60 days. And as long  as you have that timeline saying we have to see [15:52] this leading converting metric happening that we  know drives towards outcomes. And if it doesn't, [15:58] kill it. because otherwise you're wasting  time on AI that's not helping you anyways. [16:04] Okay. So, three things if you can do this week  which would be one outcome whatever role you have [16:09] marketing revops sales enablement whatever one  outcome that a CFO would recognize and actually [16:14] respect. What is that outcome that's on business  impact two is map out one workflow and I would say [16:21] pick the top three data points and top three  tasks of yours humans. That's it. Don't worry [16:26] about the 180 because that's overwhelming. top  three data points, top three tasks of a human, [16:31] and then on any AI, put a kill date of whatever  it is. It's going to change based on the impact [16:36] it's doing, but figure out is it 30 days,  is it 90 days, and measure it and hold [16:40] it accountable for that performance.  And if it doesn't make it, cut it off. [16:47] Again, think about the Aura matrix. If you're  out of place, don't feel pressure you have to [16:51] be reinventing now. If you're not there and  your team's not ready, there's nothing wrong [16:55] with getting 10 to 30% more productivity.  That's a win, just so you know. So, please [16:59] don't feel like I'm saying go out and transform if  you're not ready. If all you can do is optimize, [17:03] that's great. Focus there and then move up.  But if you're in an or that's like wanting to [17:08] do something drastic, anyone see the news from uh  Block and Jack and how he laid off 4,000 people [17:12] and the new hierarchy thing he puts out, which is  fascinating, he's blowing up his entire or at a [17:18] a 10,000 person company. This is going to change  based on everyone's organizations and your teams. [17:23] So, you just have to decide for yourself to  be comfortable with whatever level you're [17:27] on and be okay if you're on optimize and not  stress out about we need to transform because [17:32] that's not going to help you either in the long  run. And that's my time. Any questions for you? [17:45] The kill date that you're talking about, how  do you set that and what informs how you set [17:50] the kill date? It depends on what outcome you're  going for. So if you're in an enterprise motion, [17:54] you have a year and a half long sales cycle. You  have to look at the metrics that you can measure [17:58] in a quarter and figure out what's the metrics  that drive towards that year and a half sales [18:03] cycle that we can see an impact based on the  tool. So that might change if you have a like [18:07] a couple of our customers have a 5day sales  cycle so they get feedback really quick. So [18:11] you'd have to figure out where do we start to  shift that the process. Does that help? Yeah. [18:17] in your in your slide that level four maturity and  the autonomous agent is at level two performance. [18:24] Yeah. So is it because of the deficiency of uh AI  or how do you explain to the board let's say if [18:30] you set the expectation for autonomous of level  four but at the expectation at the performance [18:36] of L2 how do you justify that of the readiness of  the orgam mean? Yeah, if you go back to the slide, [18:44] I go back here. See if we can go back here.  You're talking about Hang on. Almost there. Wait, [18:50] people. Sorry. Go back. Exactly. Right. Yeah. So,  if you look at the capabilities of each level, all [18:55] of this is dictated by obviously the systems you  have in place, what data is being orchestrated, [18:59] and what the humans are doing in the meantime. All  that's got to be figured out. That doesn't just [19:03] happen in a month. So, if like they're saying  go transform with AI and it's been 30 days, [19:07] you're like, uh, we're only at level one  or two. And it all depends on did you blow [19:11] it up and you have to have that time to kind  of see if it's working or not. So this would [19:15] change based on if you're on the O, on the  A or the R because if you blow things up, [19:20] you'll you'll be faster at L4 versus somebody  who's optimizing. Does that help? Hopefully. So [19:26] you need to help them understand like you asked  us to do this AI transformation 30 days ago. [19:31] We're barely getting there because we got to get  the humans in place, the data in place, and the [19:34] process in place. and they hopefully I would hope  they'd appreciate that not just to throw AI at the [19:42] helpful. Okay. Yeah. My question is like when you  go for a AI liability versus human liability. So [19:49] if you explain to the board level let's say you  drive uh your um amplifying level of AI Mhm. But [19:57] at the same time because of some kind of you  know autonomous is not mature enough because [20:03] of the data point or any other process deficiency.  What you are saying is is it performing at a level [20:10] two instead of level four of the maturity of AI  or what is the phenomena behind that? I want to [20:17] understand. Um let's talk afterwards. We'll go  into it if that's okay because we have like 30 [20:22] seconds. Want to make sure I answer a question  too. Let's talk. Okay. Yeah. She she had one [20:31] great presentation. Um thank you. So in  your strategy piece the optimize piece that [20:37] um you to execute something like this and having  a finish date right optimize may require that [20:45] we actually look at the process and improve the  process before we even optimize anything. Right? [20:51] Would that be kind of Yes. There's going to be  times where AI is not the answer because you need [20:56] to optimize something has nothing to do with AI.  But that's part of why you want to start with the [20:59] outcomes first to figure out is this an AI thing  that to fix or not. We're talking about right have [21:08] to be changed and in order for us what we have.  Yeah. Optimize would infer that the process goes [21:16] faster. So the process itself doesn't change as  going faster versus blowing it up. Yeah. Yeah. [21:22] That takes time. So takes time putting a time  frame to something that may change may change [21:27] uh from a process perspective and data perspective  because then with different indicators may not [21:34] give us that timeline right that you were talking  about. Correct. Yeah. It's just understanding [21:39] like how long do we want to hold this process in  place to figure out is it working? Is it getting [21:43] the outcome we're trying to go for which takes  time. Yeah. Yeah. Thank you all. Appreciate it.