Eightfold Launches TalentForge Tool to Replace Packaged Enterprise HR

Why Your Next Enterprise App Won't Come From a Vendor With Eightfold AI — Ai4

Ai4

As autonomous coding erodes software development cost arbitrage, enterprises will build hyper-custom internal applications instead of buying rigid packaged SaaS from vendors.

AI-assisted summary of Why Your Next Enterprise App Won't Come From a Vendor With Eightfold AI — original source. Check the original for full context.

Source published: 2026-08-20

Detailed notes

## Background Published on YouTube by the Ai4 channel on August 20, 2026, this keynote features Ashutosh Garg, CEO of Eightfold AI. According to the page description, the session explores how collapsing software development barriers and AI advances are ending the traditional era of buying rigid packaged enterprise software, introducing Eightfold's TalentForge platform to help teams build custom applications powered by deep domain intelligence.

## TL;DR Ashutosh Garg argues that the historical SaaS model—predicated on amortizing expensive development costs across generic, one-size-fits-all software—is collapsing as AI-assisted coding drives the marginal cost of building custom software toward zero. Standardized enterprise tools force organizations into awkward compromises and multi-year vendor roadmaps, especially in human resources where rigid workflows cannot account for distinct company cultures and individual worker trajectories. To solve this, Eightfold introduced TalentForge, a natural-language application creation platform that combines billions of career data points with embedded governance and bias guardrails. By allowing employees to generate bespoke tools like tailored onboarding workflows and continuous mentorship systems in minutes, enterprises can break free from costly multi-million-dollar RFP cycles and operate as orchestrators of custom AI agents.

## Atmosphere & style The presentation carries the tone of an authoritative and strategic wake-up call, blending macroeconomic analysis of the software industry with pragmatic product vision. Garg speaks with seasoned conviction, using relatable analogies—such as the discomfort of ill-fitting off-the-rack clothing and a child's humorous definition of a CEO—to demystify complex shifts in enterprise architecture. The atmosphere feels forward-leaning and disruptive, guiding the audience from Wall Street's anxiety over a 'SaaS apocalypse' into an optimistic vision of an agile, hyper-personalized software future.

## Key takeaways - [00:04](https://www.youtube.com/watch?v=e-AzzfpT804&t=4s) Evaluating talent by backward-looking credentials and past job titles is obsolete; organizations must use deep learning to identify foundational capabilities and learning trajectory. - [02:08](https://www.youtube.com/watch?v=e-AzzfpT804&t=128s) With an estimated 70% of enterprise code now produced or assisted by AI, the collapse of software development cost arbitrage is threatening legacy SaaS business models. - [02:08](https://www.youtube.com/watch?v=e-AzzfpT804&t=128s) When the marginal cost of creating software approaches zero, enterprises no longer need to compromise on rigid, off-the-shelf packaged software. - [04:00](https://www.youtube.com/watch?v=e-AzzfpT804&t=240s) Off-the-shelf enterprise software forces distinct organizations into rigid, standardized workflows, resulting in multi-year configurations and stalled vendor feature backlogs. - [05:50](https://www.youtube.com/watch?v=e-AzzfpT804&t=350s) Generic AI coding tools fall short in HR because they lack specialized domain training on career trajectories and fail to prevent severe legal, ethical, and bias risks. - [05:50](https://www.youtube.com/watch?v=e-AzzfpT804&t=350s) TalentForge integrates deep domain intelligence with automated compliance guardrails, enabling teams to build bespoke talent software safely. - [09:05](https://www.youtube.com/watch?v=e-AzzfpT804&t=545s) Natural language application generation allows non-technical HR practitioners and team leaders to scope, prototype, and deploy custom tools in less than an hour. - [12:30](https://www.youtube.com/watch?v=e-AzzfpT804&t=750s) Knowledge workers are increasingly acting as the 'CEO of agents,' orchestrating personal automated workflows rather than executing repetitive manual tasks. - [12:30](https://www.youtube.com/watch?v=e-AzzfpT804&t=750s) Standardized onboarding fails modern enterprises because effective ramp-up workflows diverge significantly based on role, location, and seniority. - [16:49](https://www.youtube.com/watch?v=e-AzzfpT804&t=1009s) Traditional annual performance appraisals are fundamentally broken and need to be replaced by continuous, role-specific coaching and mentorship tools. - [16:49](https://www.youtube.com/watch?v=e-AzzfpT804&t=1009s) Domain-specific AI development allows enterprises to bypass multi-million dollar vendor contracts, lengthy RFP cycles, and two-year implementation roadmaps.

## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=e-AzzfpT804&t=4s) Eightfold AI and the Shift from Past Credentials to Future Potential

Eightfold AI has established itself as an enterprise foundation for talent intelligence, trusted wall-to-wall by major global corporations ranging from Vodafone and Citibank to Microsoft, Nvidia, Netflix, JPMorgan Chase, and Morgan Stanley. Between 30% and 40% of Fortune 500 companies rely on the platform across every vertical to recruit top talent, develop internal employee skills, resource critical projects, chart career paths, and transition into skills-based organizations. At the heart of this scale is a deep learning AI architecture that Eightfold pioneered well before the current wave of industry hype.

The core philosophy underlying this technology is that organizations must stop evaluating talent solely on backward-looking credentials—such as specific universities attended, past employer brand names, or historical job titles—and instead evaluate individual potential. Looking for people who have already done the exact work is practically impossible in rapidly evolving areas like enterprise AI. Success hinges on identifying individuals who possess the foundational capabilities to learn and execute quickly. Having worked in artificial intelligence for over thirty years, the current moment represents a structural pivot not just in *what* is being built, but fundamentally in *how* software itself is created.

- **Enterprise Adoption:** 30% to 40% of Fortune companies currently run Eightfold across hiring, retention, internal mobility, and skills resourcing. - **Core Methodology:** Moving away from static keyword matching and resumes to deep learning models that evaluate learning trajectory and future capability.

---

[02:08](https://www.youtube.com/watch?v=e-AzzfpT804&t=128s) The Software Cost Arbitrage Collapse and the "SaaS Apocalypse"

The software industry is undergoing an unprecedented transformation driven by automated development. Gartner recently estimated that approximately 70% of code across tech vendors and enterprises is now vibe-coded, written by AI, or produced by autonomous coding agents. This shift has triggered panic across public markets, prompting Wall Street to question whether the world is facing a "SaaS apocalypse" where packaged software ceases to exist. SaaS incumbents have suffered dramatic market contractions, with companies like HubSpot dropping as much as 70% from their peaks alongside sharp pullbacks across Workday, Salesforce, and ServiceNow.

The traditional enterprise SaaS industry was founded on a simple economic principle: the cost-of-development arbitrage. Because building software in-house was historically prohibitively expensive and technically difficult, enterprises bought packaged software from third-party vendors who could amortize development costs across thousands of customers. When the marginal cost of creating software drops toward zero, the economic rationale for purchasing rigid, shrink-wrapped applications falls apart. However, this does not spell the end of software; rather, it marks the beginning of a true "golden age of software" built around total customization.

---

[04:00](https://www.youtube.com/watch?v=e-AzzfpT804&t=240s) The Flaw of "Small, Medium, Large" Enterprise Software

To illustrate why packaged enterprise software is failing modern organizations, shopping for clothes serves as a telling analogy: almost nobody enjoys waiting outside fitting rooms trying on off-the-rack garments sold in rigid sizes like Small, Medium, or Large. Human beings come in every shape and size, leaving people with closets full of clothes that do not quite fit and must be uncomfortably stretched. Enterprise software has forced companies into the exact same compromise for decades.

```mermaid flowchart TD accTitle: Legacy SaaS vs Custom AI Enterprise Software subgraph Legacy["Traditional SaaS Dilemma"] A["Standardized App (S/M/L)"] --> B["Multi-Year Configuration"] B --> C["Compromised Workflows"] C --> D["2-Year Change Request Roadmaps"] end subgraph Future["Agentic Domain-Specific Development"] E["Domain Intelligence Engine"] --> F["Natural Language Prompting"] F --> G["Instant Custom Software"] G --> H["Dynamic, Team-Specific Evolution"] end ```

A major bank like Citibank is fundamentally different from HSBC—they possess different workplace cultures, operating procedures, risk appetites, and technology stacks. Yet enterprise vendors force them to purchase identical, off-the-shelf HRIS, CRM, and financial planning platforms. Organizations spend years and millions of dollars heavily configuring these generic tools, only to submit change requests that sit in vendor product backlogs for one to two years whenever a workflow does not fit the business.

---

[05:50](https://www.youtube.com/watch?v=e-AzzfpT804&t=350s) Introducing TalentForge: Claude Code for the HR Domain

Eightfold is redefining enterprise software with **TalentForge**, a platform designed to let organizations, managers, and employees build custom-tailored talent applications for themselves. While general-purpose development assistants like Claude Code, OpenAI Codex, Vercel, or Lovable can generate code, they lack the specialized foundation necessary to operate in the talent domain. Building enterprise HR applications requires two non-negotiable pillars: deep domain intelligence and embedded compliance trust.

```mermaid flowchart LR accTitle: TalentForge Architectural Trust and Intelligence Pipeline subgraph DomainData["Eightfold Domain Foundation"] A["Billions of Career Trajectories"] B["Skills & Capability Graphs"] end subgraph Guardrails["Autonomous Trust Engine"] C["Bias & Ethics Mitigation"] D["Labor Law Compliance"] E["Data Privacy Guardrails"] end subgraph Output["Custom Applications"] F["Tailored Recruiting"] G["Adaptive Onboarding"] H["Continuous Mentorship"] end

DomainData --> Guardrails Guardrails --> Output ```

Domain intelligence cannot be substituted with generic LLMs; it requires models trained on billions of real-world career trajectories and data points to understand actual skill adjacency, potential, and workplace success without falling back on naive keyword searches. Furthermore, HR workflows carry massive legal, ethical, and governance risks. A generic AI agent might mistakenly attempt to filter job candidates by age, trigger automated rejection emails whenever interviewers submit negative feedback, or prompt applicants for invasive personal details like home addresses and family members. TalentForge embeds governance, compliance, and anti-bias guardrails directly into the generation pipeline so custom applications remain legally sound and ethically compliant.

---

[09:05](https://www.youtube.com/watch?v=e-AzzfpT804&t=545s) The Natural Language App Creation Workflow

Because Eightfold is already deployed across approximately 70% of trillion-dollar enterprises, top banks, healthcare networks, and pharmaceutical giants, organizations already have access to the underlying infrastructure required to build bespoke applications. Rather than waiting on centralized IT or third-party vendors, individual team leaders, employees, and HR practitioners can create tools directly via a no-code developer dashboard using natural language prompts.

```mermaid flowchart TD accTitle: TalentForge Natural Language Build Lifecycle A["Describe Application in Plain English (e.g., Team Skills Matrix, HRIS)"] --> B["System Guided Scoping & Clarification"] B --> C["Autonomous Requirements Engine (Functional & Compliance/Bias Checks)"] C --> D["Interactive Prototype Visualization"] D --> E["One-Click Test & Enterprise Deployment"] ```

The build process translates simple text descriptions into enterprise-ready software within an hour: 1. **Describe the Intent:** The user types out what they need in plain English—such as a tailored performance management tool, an organizational analytics view, or a skill-development tracker. To prove this ease of use, a junior HR employee at Eightfold built a complete, working HRIS independently in under a week. 2. **Clarification and Scoping:** The platform asks follow-up questions to refine specific nuances and operational requirements. 3. **Automated Requirements Generation:** The engine synthesizes functional specifications while automatically enforcing non-functional requirements around compliance, governance, data privacy, bias prevention, and scalability. 4. **Interactive Visualization:** The builder reviews and modifies a live visual interface before committing. 5. **Testing and Publishing:** The application is tested, validated, and deployed into production in a single workflow.

---

[12:30](https://www.youtube.com/watch?v=e-AzzfpT804&t=750s) Becoming the "CEO of Agents" and Personalizing Workflows

As professionals advance in their careers, their daily responsibilities shift toward communication, alignment, and coordination. When asked what a CEO does, young children often observe the reality accurately: "He talks and talks, makes a phone call, goes to a meeting and talks, and comes home and makes another phone call." This highlights a fundamental truth about modern knowledge work: professionals accomplish their work through other entities. In the AI era, every employee is evolving to become the "CEO of agents," orchestrating personal workflows and requiring custom software to execute them effectively.

Onboarding highlights the failure of one-size-fits-all software. In typical enterprises, getting a new hire fully productive takes three months. A standardized onboarding flow cannot account for the fact that a software engineer needs a completely different ramp-up experience than a financial analyst or an HR recruiter. Even within the same role, an engineer hired in Mountain View has different logistical and operational contexts than one hired in London, just as a 15-year veteran requires a different ramp than a fresh college graduate. While traditional SaaS vendors spent decades building rigid, standalone onboarding tools, TalentForge allows teams to generate tailored, hyper-personalized onboarding workflows in minutes.

---

[16:49](https://www.youtube.com/watch?v=e-AzzfpT804&t=1009s) Reclaiming Agility from the Multi-Million Dollar RFP Cycle

Traditional performance reviews represent another fundamentally broken enterprise category. Standard appraisal systems are backward-looking mechanisms built to penalize employees and justify compensation decisions rather than foster development. Managers dread difficult rating reviews, and employees dread receiving them. Modern teams require continuous mentorship and coaching tools that reflect role-specific realities—an account executive cannot be evaluated or developed through the same framework as a software engineer, product manager, or content writer.

Global enterprises are already using TalentForge to break out of legacy cycles: - **Industry Adoption:** Major international telecommunications providers and global professional services firms like Deloitte are actively building internal applications using TalentForge. - **Managerial Visibility:** Managers can instantly generate custom dashboards to monitor operational rhythms, such as tracking which direct reports are conducting regular one-on-ones, preventing team blind spots without manual overhead. - **RFP Disruption:** Organizations no longer need to spend half a million dollars, run multi-year RFP cycles, and endure two-year implementation timelines to deploy basic management tools.

Because neither human beings nor modern enterprises fit into standardized Small, Medium, and Large boxes, enterprise computing must move toward custom-built, agent-driven software. The presentation concluded with an invitation for attendees to see live demonstrations of TalentForge at Booth 1354.

## Notable quotes > For twenty years, buying enterprise software meant filing requirements, waiting on a vendor's roadmap, and living with a workflow no one on the team actually chose. — Ashutosh Garg

> The next wave of enterprise applications will come from the people who understand the problem best, armed with real domain intelligence instead of a generic model. — Ashutosh Garg

> instead of selling you the finished app, we're handing you the domain expertise and the tools to build the one your team actually needs. — Ashutosh Garg

> [00:04](https://www.youtube.com/watch?v=e-AzzfpT804&t=4s) Looking for people who have already done the exact work is practically impossible in rapidly evolving areas like enterprise AI. Success hinges on identifying individuals who possess the foundational capabilities to learn and execute quickly. — Ashutosh Garg

> [02:08](https://www.youtube.com/watch?v=e-AzzfpT804&t=128s) When the marginal cost of creating software drops toward zero, the economic rationale for purchasing rigid, shrink-wrapped applications falls apart. — Ashutosh Garg

> [04:00](https://www.youtube.com/watch?v=e-AzzfpT804&t=240s) almost nobody enjoys waiting outside fitting rooms trying on off-the-rack garments sold in rigid sizes like Small, Medium, or Large. Human beings come in every shape and size, leaving people with closets full of clothes that do not quite fit and must be uncomfortably stretched. Enterprise software has forced companies into the exact same compromise for decades. — Ashutosh Garg

> [05:50](https://www.youtube.com/watch?v=e-AzzfpT804&t=350s) While general-purpose development assistants like Claude Code, OpenAI Codex, Vercel, or Lovable can generate code, they lack the specialized foundation necessary to operate in the talent domain. — Ashutosh Garg

> [12:30](https://www.youtube.com/watch?v=e-AzzfpT804&t=750s) He talks and talks, makes a phone call, goes to a meeting and talks, and comes home and makes another phone call. — Ashutosh Garg

> [12:30](https://www.youtube.com/watch?v=e-AzzfpT804&t=750s) In the AI era, every employee is evolving to become the 'CEO of agents,' orchestrating personal workflows and requiring custom software to execute them effectively. — Ashutosh Garg

> [16:49](https://www.youtube.com/watch?v=e-AzzfpT804&t=1009s) Standard appraisal systems are backward-looking mechanisms built to penalize employees and justify compensation decisions rather than foster development. — Ashutosh Garg

## Mentioned Ashutosh Garg, Eightfold AI, TalentForge, Vodafone, Citibank, Microsoft, Nvidia, Netflix, JPMorgan Chase, Morgan Stanley, Gartner, HubSpot, Workday, Salesforce, ServiceNow, HSBC, Claude Code, OpenAI Codex, Vercel, Lovable, Deloitte, Ai4

## Video description Ashutosh Garg - CEO, Eightfold AI

For twenty years, buying enterprise software meant filing requirements, waiting on a vendor's roadmap, and living with a workflow no one on the team actually chose. That era is ending: the barriers to building custom software are collapsing right alongside AI's own advances, and even experienced engineers now build with AI as their default toolkit, not the exception. The next wave of enterprise applications will come from the people who understand the problem best, armed with real domain intelligence instead of a generic model.

Eightfold's TalentForge — a platform that lets teams build custom talent applications on top of deep, purpose-built domain intelligence rather than a generic LLM — brings this to life. It's a different role for a vendor to play: instead of selling you the finished app, we're handing you the domain expertise and the tools to build the one your team actually needs. Attendees will leave with a sharper framework for rethinking build versus buy, and a clear-eyed look at what it actually takes to turn domain expertise into working software. For anyone setting technology strategy for their organization, this is a shift worth getting ahead of before competitors do.

[Source](https://www.youtube.com/watch?v=e-AzzfpT804)

中文简报

Eightfold推出TalentForge，拟取代企业HR套装软件随着自主编程削弱软件开发的成本套利优势，企业将构建高度定制的内部应用，而非向供应商采购僵化的SaaS套装软件。 Ashutosh Garg认为，传统的SaaS模式建立在通过通用的“一刀切”软件摊销昂贵开发成本的基础之上，而随着AI辅助编程将构建定制软件的边际成本推向趋近于零，该模式正在崩溃。标准化企业工具迫使组织做出尴尬的妥协并受制于长达数年的供应商开发路线图，人力资源领域尤甚，其僵化的工作流无法适应差异化的企业文化与员工个人发展轨迹。为此，Eightfold推出了TalentForge，这是一个基于自然语言的应用构建平台，将数十亿职业数据点与内置治理及防偏见机制相结合。通过让员工在数分钟内生成如定制化入职流程和持续导师系统等专属工具，企业得以摆脱耗资数百万美元的RFP采购周期，转型成为定制AI代理的协调者。依据过往资历和以往职位头衔评估人才已经过时；组织必须利用深度学习来识别员工的核心基础能力与学习成长轨迹。 (0:04) 据估计目前已有70%的企业代码由AI生成或辅助完成，软件开发成本套利优势的瓦解正对传统SaaS商业模式构成威胁。 (2:08) 当构建软件的边际成本趋近于零时，企业不再需要在僵化的现成套装软件上做出妥协。 (2:08) 现成的企业软件迫使不同组织削足适履地套用僵化、标准化的工作流，导致配置耗时数年且供应商功能储备积压停滞。 (4:00) 通用AI编程工具在人力资源领域表现不足，因其缺乏针对职业轨迹的专业领域训练，且无法防范严重的法律、道德和偏见风险。 (5:50)

Reading summary

As automated code generation reduces software development costs toward zero, the economic rationale underpinning packaged enterprise SaaS is rapidly eroding. Eightfold AI argues that organizations will replace rigid, third-party software with custom, domain-specific applications built internally through natural language prompts and automated compliance guardrails. The Collapse of SaaS Cost Arbitrage (2:08) Traditional enterprise software vendors relied on a cost-of-development arbitrage: building in-house software was prohibitively expensive, allowing third parties to amortize development costs across thousands of corporate buyers. That model is destabilizing as development costs collapse. Gartner estimates that approximately 70% of code across enterprises and tech vendors is now produced by AI, vibe-coding, or autonomous coding agents. This shift has rattled public markets, contributing to steep contractions among SaaS incumbents, including drops of up to 70% from peaks for HubSpot alongside sharp pullbacks for Workday, Salesforce, and ServiceNow. The Limits of Standardized Software (4:00) Packaged enterprise applications force disparate companies into standardized, off-the-rack templates. Major financial institutions such as Citibank and HSBC operate with distinct internal cultures, regulatory risk appetites, and IT architectures, yet enterprise vendors sell them identical HRIS, CRM, and planning systems. Adapting these rigid platforms routinely demands millions of dollars and multi-year implementation cycles, while subsequent operational adjustments remain stalled in vendor feature backlogs for one to two years. Domain Intelligence and Compliance Guardrails (5:50) General-purpose coding assistants like Claude Code, OpenAI Codex, Vercel, and Lovable can generate functional code, but Eightfold argues they lack the domain intelligence and regulatory guardrails required for enterprise human resources. Talent software cannot rely on naive keyword matching; it requires models trained on real career trajectories and skill adjacencies to assess individual potential over static resumes. Furthermore, unconstrained AI agents introduce severe legal, ethical, and governance liabilities. Without embedded guardrails, automated agents can inadvertently filter candidates by age, solicit invasive personal data, or trigger unlawful rejections. To address this, Eightfold designed TalentForge to enforce labor law compliance, bias mitigation, and data privacy directly within the application generation pipeline. Natural Language Development and Bespoke Workflows (9:05) TalentForge lets HR personnel, team leads, and employees build custom software via plain-English prompts rather than waiting on centralized IT or running half-million-dollar RFP cycles. The system clarifies functional requirements, generates compliance specifications, renders an interactive prototype, and deploys the application in hours. To illustrate feasibility, Eightfold reported that an internal junior HR employee built an operational HRIS in under a week. This architecture targets workflows where standardized software regularly underperforms, such as onboarding and performance tracking. Because ramp-up requirements vary widely by seniority, location, and role, teams can configure hyper-specific onboarding flows or continuous coaching dashboards in minutes. Enterprise adopters, including Deloitte and international telecommunications carriers, are already using the platform to deploy custom managerial tools without multi-year procurement delays. Eightfold contends that enterprise computing is entering a custom-built, agent-driven era where employees act as orchestrators of bespoke software rather than users of packaged applications. The broader viability of this shift hinges on whether domain-specific platforms can reliably maintain regulatory compliance and governance as non-technical employees assume the role of software creators.

精炼阅读摘要

Eightfold AI 指出，随着生成式 AI 和自动化编程使软件边际开发成本趋于归零，传统标准化企业 SaaS 的成本套利模型正在瓦解；未来的企业应用将不再依赖外部供应商购买通用套件，而是转向依托垂直领域智能与内置合规引擎、通过自然语言自主构建完全定制化的专属应用。开发成本骤降打破传统 SaaS 商业逻辑 (2:08) 软件行业正在经历由自动化编程引发的结构性重塑。根据 Gartner 的统计预估，目前科技厂商与企业内部约有 70% 的代码已通过 AI 协作（如 vibe-coding）或自主代码智能体编写。这一转变打破了传统企业 SaaS 赖以生存的基础——即“开发成本套利”模式。以往由于企业自研软件成本高昂且技术门槛极高，只能向第三方供应商购买打包软件，由后者将研发费用分摊给成千上万家客户。当边际代码编写成本迅速降低，购买僵化盒装软件的经济合理性不复存在。资本市场已对此产生强烈震荡并引发“SaaS 末日”的担忧，HubSpot 股价自高点下跌多达 70%，Workday、Salesforce 和 ServiceNow 等传统巨头也遭遇显著回调。Eightfold 认为这并非软件的终结，而是完全客制化软件“黄金时代”的开端。标码化套件与企业个性化现实的脱节 (4:00) 传统企业级软件往往将所有客户塞进固定规格的框架中，如同商场里缺乏弹性的“小号、中号、大号（S/M/L）”成衣。不同企业之间存在巨大的根本差异，例如同为全球顶尖银行的花旗银行（Citibank）与汇丰银行（HSBC），在组织文化、操作流程、风险偏好和底层技术栈上截然不同，传统模式却迫使它们采购功能完全同质的 HRIS、CRM 或财务系统。为了弥合这种差距，企业往往要耗费数年时间和数百万美元对通用工具进行重度二次配置；一旦出现新的业务流程诉求，修改请求通常会在供应商的产品排期待办（backlog）中积压一到两年。此外，传统体系依赖过去的静态标签（如毕业院校、历史职位与前雇主品牌）筛选人才，已无法适应企业 AI 时代对学习潜力和底层综合能力的敏捷评估需求。垂直领域智能与合规护栏的双重壁垒 (5:50) 为了替代外购通用软件，Eightfold 推出了专用于人才领域的低门槛开发工具 TalentForge。演讲者强调，虽然 Claude Code、OpenAI Codex 或 Lovable 等通用代码生成工具层出不穷，但构建企业级人力资源系统必须具备两大前提：深度的行业领域智能和内嵌合规信任。在领域智能方面，模型需要理解技能关联性与职业发展路径，不能退化为简陋的简历关键词匹配；这依赖于 Eightfold 训练的数十亿真实职业轨迹与技能图谱。在合规与治理方面，人力资源业务蕴含极高的法律与伦理风险，未经校准的通用智能体可能违规按年龄过滤候选人、在面试官提交负面反馈后自动触发拒信，或索取过度侵犯隐私的个人信息。TalentForge 将劳动法合规、防偏见和数据隐私护栏直接固化到底层生成流程中，确保自动构建的应用合法合规。自然语言构建流程与智能体工作流编排 (9:05) 借助已在众多财富 500 强企业中铺设的基础设施，业务主管、HR 从业者乃至普通员工均可通过无代码仪表盘，以自然语言直接创建工具。整个流程包括描述意图、系统交互厘清细节、自动生成包含合规与隐私验证的功能规范、交互式原型预览，以及一键测试与生产部署；Eightfold 内部初级 HR 员工便曾在一周内独立搭建出完整的可用 HRIS。随着知识工作者日益演变为“智能体 CEO”（CEO of agents），每个人都需要通过调度专属智能体来推进工作流。以新员工入职（Onboarding）为例，传统入职流程通常耗时三个月且高度固化，难以兼顾不同职能、不同工作地点（如山景城与伦敦）以及资深员工和应届毕业生的现实差异，而自主构建平台可以在几分钟内生成针对具体岗位的超个性化入职路径。颠覆繁重采购周期并重构管理实践 (16:49) 以绩效评估为代表的传统企业管理模块长期处于失灵状态：传统的年度考核本质上是事后追责、用于佐证薪酬决策的防御性机制，令管理者与下属皆备受困扰。现代团队所需要的是持续性、因岗位而异的教练式辅导工具，例如销售代表与软件工程师的发展评估框架必然迥异。目前，包括德勤（Deloitte）和大型跨国电信运营商在内的企业已在利用 TalentForge 摆脱旧有循环。管理者能即时生成个性化仪表盘来监督团队节奏（例如跟踪下属的一对一沟通频率），消除管理盲区。这使企业彻底免于发起耗资数十万美元、耗时数年的漫长 RFP（征求方案书）竞标与实施周期，真正重获组织的管理敏捷度。 Eightfold 明确宣告，由于现代企业及其员工的复杂异构性无法被标准化软件套件所容纳，企业级计算正在加速转向基于领域智能体的高度定制模式；然而，这种由一线自主构建工具的分布式范式在大规模企业环境中能否长期维持系统架构的一致性与治理稳定性，仍是值得持续观察的议题。

Detailed notes — zh

## Background 本内容由 Ai4 频道于 2026 年 8 月 20 日发布在 YouTube 平台。视频由 Eightfold AI 首席执行官 Ashutosh Garg 主讲，主要探讨了企业软件采购时代的终结、AI 赋能下自研定制软件门槛的崩塌，以及如何借助 TalentForge 等具备深厚领域智能的平台构建真正贴合企业实际需求的专属人才应用。 ## TL;DR Ashutosh Garg 指出，随着 AI 自动化编写代码的普及，传统 SaaS 依赖多租户分摊高昂研发成本的“开发成本套利”模式已走向瓦解。过去企业被迫购买标准化的成品软件并经历漫长的定制周期，如同穿上不合身的标准尺码成衣；而现在软件正在进入人人可定制的“黄金时代”。为此，Eightfold 推出了专为 HR 领域打造的应用构建平台 TalentForge，它融合了数十亿级职业数据与原生合规护栏，让非技术员工仅凭自然语言提示词即可在数小时至一周内自主构建出入职培训、教练型绩效评估等专属合规企业应用。 ## Atmosphere & style 演讲现场充满自信与前瞻性，主讲人以风趣幽默的生活化类比（如逛街试衣间、向孩子解释工作）打破严肃的技术探讨，迅速拉近与观众的距离。整体叙事节奏紧凑有力，不仅直击企业采购软件中的核心痛点，更以极具感染力的愿景向台下传达了对软件自主构建时代的坚定信念，带给听众一种颠覆传统工作流、技术真正赋能个体的启发感与兴奋感。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=e-AzzfpT804&t=4s) Eightfold 在财富 500 强企业中渗透率达 30% 至 40%，其核心理念是将人才评估从静态的历史履历筛选转向基于深度学习预测未来潜能与学习敏锐度。 - [02:08](https://www.youtube.com/watch?v=e-AzzfpT804&t=128s) 随着企业中 70% 的代码由 AI 生成或辅助编写，传统 SaaS 赖以生存的“研发成本分摊与开发成本套利”商业逻辑正在彻底瓦解。 - [04:00](https://www.youtube.com/watch?v=e-AzzfpT804&t=240s) 传统企业软件如同标准尺码成衣，强迫业务流程千差万别的企业削足适履，陷入漫长的定制配置与等待供应商路线图的死循环。 - [05:50](https://www.youtube.com/watch?v=e-AzzfpT804&t=350s) Eightfold 推出面向 HR 领域的定制构建平台 TalentForge，让企业能够基于深层人才领域智能自主构建个性化应用，而非直接购买通用成品套件。 - [07:20](https://www.youtube.com/watch?v=e-AzzfpT804&t=440s) 通用代码生成工具在人力资源场景存在显著短板，缺乏沉淀自数十亿职业数据的人才洞察，且极易踩中偏见与合规伦理红线。 - [09:33](https://www.youtube.com/watch?v=e-AzzfpT804&t=573s) TalentForge 支持纯自然语言零代码交互，能自动推导补齐数据合规与防偏见等非功能性需求，将应用交付周期从数月缩短至一小时。 - [12:30](https://www.youtube.com/watch?v=e-AzzfpT804&t=750s) 现代职场人正逐渐转变为“管理众多 AI Agent 的 CEO”，这种角色转变对高度个性化、定制化的数字基础设施提出了刚性需求。 - [14:10](https://www.youtube.com/watch?v=e-AzzfpT804&t=850s) 个性化入职培训与向前看的成长型绩效辅导长期受困于标准化软件，企业亟需根据岗位与阶段量身定制工具以替代传统的向后考核机制。 - [16:49](https://www.youtube.com/watch?v=e-AzzfpT804&t=1009s) 自建软件门槛的大幅下降正在终结耗费数十万美元与数年周期的传统招标采购模式，企业软件迎来由一线业务人员主导构建的黄金时代。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=e-AzzfpT804&t=4s) Eightfold 的背景与 AI 驱动的人才潜力评估作为 Eightfold AI 的首席执行官，Ashutosh Garg 在开场时介绍了公司的定位与市场影响力。Eightfold 深受全球各行业头部企业的信赖，包括 Vodafone、花旗银行（Citi），以及微软（Microsoft）、英伟达（NVIDIA）、奈飞（Netflix）、摩根大通（JPMorgan Chase）和摩根士丹利（Morgan Stanley）等巨头。目前，财富 500 强企业中有近 30% 到 40% 的行业领导者都在全面（wall-to-wall）使用 Eightfold，覆盖从精准招聘、员工技能培训、项目资源调配、内部流动性支持到职业发展路径规划和技能型组织转型的全流程。支撑这一庞大规模的是 Eightfold 多年来开创的深度学习技术。Ashutosh 指出，传统评估人才的方式往往局限于“他们是谁、过去在哪里工作过、拥有什么学位”，而 Eightfold 从创立之初就致力于打破这种局限，转向关注每个人的“未来潜能”——即“重点不在于他们过去做过什么，而在于他们下一步能做什么”。他提到，在如今漫步 AI 展厅、探索深度学习落地的背景下，如果企业依然执着于寻找已经精通 AI 的现成人才，几乎是不可能完成的任务；真正的出路在于发掘那些具备快速学习和实践能力的人才。他本人在 AI 领域深耕了 30 年，早在当前的 AI 热潮之前就坚信这一技术方向。 - **财富 500 强渗透率**：全球顶级银行、电信运营商、医疗健康与制药龙头中，30% 至 40% 的企业均采用 Eightfold 作为核心人才平台。 - **人才评估核心理念**：从“基于历史履历与学历的静态筛选”转变为“基于深度学习预测个人未来潜能与学习敏锐度”。 --- ### [02:08](https://www.youtube.com/watch?v=e-AzzfpT804&t=128s) 软件开发成本套利的终结与 SaaS 的未来当前全球正在经历一场截然不同的技术变革，其核心区别不再是“我们在构建什么”，而是“我们如何构建软件”。Gartner 最近发布的数据显示，企业或软件供应商编写的代码中，已有 70% 是由 AI 生成、通过 AI Agent 辅助或借助 v0/AI 编程工具完成的。这引发了外界对于“SaaS 终结/大灾难（SaaS Apocalypse）”的恐慌：传统软件是否将不复存在？观察公开市场，诸如 Workday、Salesforce、ServiceNow 等 SaaS 标杆企业的估值与股价均承受巨大压力，而像 HubSpot 这样的公司股价跌幅甚至达到了 70%。针对这种市场恐慌，Ashutosh 剖析了软件产业的底层经济模型：过去的整个企业级软件生态都建立在“开发成本套利（cost of development arbitrage）”的基础之上。因为过去自研软件门槛太高、成本过于昂贵，企业只能退而求其次，向外部供应商购买标准化的成品软件，让供应商在成百上千家客户之间分摊研发成本。然而，当 AI 让代码构建成本断崖式下跌时，继续购买通用、封装好的成品软件（shrink-wrapped software）的逻辑便彻底瓦解。但他与华尔街的悲观预期相反，他坚定地认为，世界正在步入“软件的黄金时代”。 - **行业开发范式转移**：企业与供应商 70% 的代码已依赖 AI 自动化编写或辅助生成。 - **商业模式瓦解**：传统 SaaS 依赖的“多租户研发成本分摊”优势，正在被极低的代码自建成本所打破。 --- ### [04:00](https://www.youtube.com/watch?v=e-AzzfpT804&t=240s) 试衣间隐喻与“削足适履”的传统企业软件为了解释为何这标志着黄金时代的到来，Ashutosh 向现场观众抛出了一个生活化的类比：“你们有多少人真心喜欢逛街买衣服？又有谁喜欢站在试衣间外面排队试衣服？”面对台下几乎无人举手的反应，他笑着说道，成衣市场通常只提供 S、M、L（小号、中号、大号）三种标准尺码，但现实中人类的身材体型千差万别，结果就是每个人的衣柜里塞满了“看起来还行但并不真正合身、只能勉强拉扯着穿”的衣服。这种尺码错配与当前的企业软件困境如出一辙。花旗银行之所以成为花旗银行，绝不是因为它的运作流程和汇丰银行（HSBC）完全一样；不同企业的组织文化、员工构成与底层技术栈各不相同。然而在过去的几十年里，所有企业却被强制购买完全相同的标准 CRM、HRIS 或财务规划系统，随后耗费数年时间去进行痛苦的定制配置。“每一次你觉得软件不顺手，就必须提交变更需求（Change Request），然后陷入等待一到两年的供应商路线图排期中。”Ashutosh 强调，这种被迫削足适履的恶性循环已经到了必须改变的历史节点。 --- ### [05:50](https://www.youtube.com/watch?v=e-AzzfpT804&t=350s) 推出 TalentForge：面向 HR 领域的定制化构建平台为了彻底重构这一软件采购范式，Eightfold 推出了 TalentForge 平台，旨在让企业能够完全自主地构建高度定制化的人才与人力资源应用。为了便于技术受众理解，Ashutosh 将其形象地概括为“HR 领域的 Claude Code”。 ```mermaid flowchart TD accTitle: TalentForge Architecture Flow accDescription: Flow chart showing user input transformed via TalentForge domain intelligence and trust guardrails into enterprise custom applications. subgraph User["用户层 (User Layer)"] UI["自然语言描述业务需求 (Natural Language Prompt)"] end subgraph Core["TalentForge 核心引擎 (Domain Engine)"] DI["数十亿级职业数据与智能 Agent (Billions of Career Data Points)"] TG["内置合规与治理护栏 (Compliance, Bias & Ethics Guardrails)"] FR["需求自动生成与校验引擎 (Functional & Non-Functional Specs)"] end subgraph Output["交付成果 (Custom Enterprise Apps)"] App1["定制化入职体验系统"] App2["成长型绩效教练看板"] App3["专属 HRIS / 技能管理应用"] end UI --> FR DI --> FR TG --> FR FR --> App1 FR --> App2 FR --> App3 ``` 面对“为何不直接使用通用代码生成工具（如 Claude Code、Codex、Vercel 或 Lovable）”的疑问，Ashutosh 指出了通用大模型在企业人才领域的两大核心短板： 1. **领域智能与人才 Agent 底座**：平台底层注入了 Eightfold 多年沉淀的数十亿级人才数据点与职业发展轨迹模型，能够深度理解谁表现优异、谁匹配特定组织、谁能快速掌握新技能，使应用不再依赖死板的关键字搜索，而是透视人才的实际潜能。 2. **信任、合规与治理护栏（Trust & Compliance）**：在 HR 场景中，模型信任与数据治理关乎法规与个人职业生涯。通用构建极易踩雷，例如开发者可能会错误地把“申请人年龄”纳入筛选维度、设置“面试全被拒则自动发拒信”的非合规全自动工作流，或让面试 Agent 违规索要家庭住址和家庭成员隐私。TalentForge 将防偏见、伦理规范、合规审计等治理机制原生嵌入底层，使非技术人员在自由构建的同时免除违规风险。 --- ### [09:33](https://www.youtube.com/watch?v=e-AzzfpT804&t=573s) 从自然语言到合规应用：TalentForge 的极简构建流程在具体使用上，TalentForge 提供了极其简便的开发者看板，任何人无需具备编程技能即可上手。用户只需使用日常语言描述想要的应用（例如：“我想要一个绩效管理系统”、“我需要一个分析看板”，或“我希望有一个管理者工作流来掌握团队技能与成长规划”），系统便会展开智能对话，引导用户完善细节。Ashutosh 透露了一个真实案例：Eightfold 内部一位年轻的初级 HR 员工，仅靠自然语言交互就在短短一周内独立构建出了一整套完整的企业级 HRIS 系统。描述完成后，平台会自动输出功能性需求，并同步补齐至关重要的非功能性需求——包括合规审查、系统扩展性、访问治理、偏见控制及伦理标准等。系统会主动提示潜在的合规风险并提供调整建议，随后自动生成应用界面的可视化预览。用户确认或修改需求后，平台将一键完成部署、测试与发布流程。原本需要数月甚至数年外采交付的应用，普通员工在短短一小时内即可构建并上线。 - **零代码自然语言交互**：通过纯文本提示词直接生成前端、后端逻辑与数据流。 - **自动推导非功能性规范**：针对 HR 领域强制注入伦理、隐私、合规与抗偏见策略。 - **极速构建周期**：常规团队级轻量应用仅需一小时即可完成从构思、调试到发布的全流程。 --- ### [12:30](https://www.youtube.com/watch?v=e-AzzfpT804&t=750s) 每个人都是“AI Agent 的 CEO”与场景定制化变革 Ashutosh 分享了一段幽默的家庭趣事：他有两个分别 9 岁和 12 岁的儿子，如果有人问孩子们“爸爸每天在公司做什么”，孩子们绝不会说“爸爸在经营公司、管理 GTM 市场流程、制定产品战略或抓销售”，而是会回答：“他就是一直在说话、说话，然后打电话继续说话，接着去开会说话，回到家又拿起电话一直说个不停。”这生动反映了现代职场人的工作本质——随着层级提升，每个人都是在通过协调人来达成目标。他指出，未来职场中每个人的角色都在演变为“管理众多 AI Agent 的 CEO”，这就要求每个人都能拥有专属于自己的个性化软件与数字基础设施。针对具体的业务场景，Ashutosh 重点剖析了两个长期被传统软件束缚的领域： - **个性化入职培训（Onboarding）**：传统企业中新员工平均需要 3 个月才能完全形成生产力，且不同职能（软件工程师 vs. 财务分析师 vs. HR 招聘官）、不同工作地（山景城 vs. 伦敦）以及不同资历（15 年经验行业老兵 vs. 应届毕业生）的入职诉求完全不同。过去有大量软件公司花费数十年仅做标准化入职系统，而现在借助 TalentForge，管理者可以在极短时间内为不同团队和个人生成专属的定制入职流程。 - **重构绩效评估（Performance Review）**：谈及绩效评估时，Ashutosh 调侃现场“没有任何人会喜欢做绩效评估，更没人愿意接受那种令人痛苦的谈话”。原因在于传统方案极其落后，充斥着向后看（backward-looking）的惩罚与定薪思维；而现代团队真正需要的是向前看（forward-looking）的教练指导与职业成长规划。面对客户经理、软件工程师、产品经理或内容文案等完全不同的工种，企业必须拥有量身定制的绩效辅导工具。目前，全球顶尖电信巨头以及德勤（Deloitte）等大型专业服务机构已经在利用 TalentForge 构建这类下一代应用。 --- ### [16:49](https://www.youtube.com/watch?v=e-AzzfpT804&t=1009s) 告别漫长采购周期，迎来软件定制的黄金时代在传统模式下，如果一线管理者想要在每周一早晨获得一份专属的团队视图（例如清晰掌握直属下级中谁完成了 1-on-1 沟通），唯一的路径就是向公司提出需求，经历繁冗的招标（RFP）流程，花费 50 万美元采购外部软件，再耗费两年的实施周期。这种高昂的成本与漫长的等待使得大多数个性化需求被直接扼杀。 Ashutosh 总结道，人类从来没有标准的小号、中号或大号身材，企业同样没有标准尺码；正是这种独特的组织差异让商业世界变得丰富多彩与充满活力。随着构建门槛的彻底瓦解，企业软件不再需要向外部供应商的标准套件妥协，未来的应用将由最懂业务痛点的企业自身和一线员工亲手打造。在演讲最后，他热情邀请参会者前往 1354 号展位亲自体验 TalentForge 的实时构建演示。 - **传统采购 vs. 自主构建**： - **传统模式**：发布 RFP $\rightarrow$ 耗资 50 万美元 $\rightarrow$ 采购耗时 2 年 $\rightarrow$ 实施配置耗时 2 年 $\rightarrow$ 获得无法完全贴合需求的通用软件。 - **TalentForge 模式**：一线人员按需定义 $\rightarrow$ 自然语言驱动 $\rightarrow$ 原生内置领域合规 $\rightarrow$ 1 小时至 1 周内极速交付专属应用。 ## Notable quotes > [00:04](https://www.youtube.com/watch?v=e-AzzfpT804&t=4s) 重点不在于他们过去做过什么，而在于他们下一步能做什么 — Ashutosh Garg > [02:08](https://www.youtube.com/watch?v=e-AzzfpT804&t=128s) 过去的整个企业级软件生态都建立在‘开发成本套利’的基础之上 — Ashutosh Garg > [04:00](https://www.youtube.com/watch?v=e-AzzfpT804&t=240s) 你们有多少人真心喜欢逛街买衣服？又有谁喜欢站在试衣间外面排队试衣服？ — Ashutosh Garg > [04:00](https://www.youtube.com/watch?v=e-AzzfpT804&t=240s) 每一次你觉得软件不顺手，就必须提交变更需求，然后陷入等待一到两年的供应商路线图排期中 — Ashutosh Garg > [05:50](https://www.youtube.com/watch?v=e-AzzfpT804&t=350s) HR 领域的 Claude Code — Ashutosh Garg > [12:30](https://www.youtube.com/watch?v=e-AzzfpT804&t=750s) 他就是一直在说话、说话，然后打电话继续说话，接着去开会说话，回到家又拿起电话一直说个不停 — Ashutosh Garg > [12:30](https://www.youtube.com/watch?v=e-AzzfpT804&t=750s) 管理众多 AI Agent 的 CEO — Ashutosh Garg > [16:49](https://www.youtube.com/watch?v=e-AzzfpT804&t=1009s) 人类从来没有标准的小号、中号或大号身材，企业同样没有标准尺码 — Ashutosh Garg ## Mentioned Ashutosh Garg, Eightfold AI, TalentForge, Vodafone, 花旗银行 (Citi), 微软 (Microsoft), 英伟达 (NVIDIA), 奈飞 (Netflix), 摩根大通 (JPMorgan Chase), 摩根士丹利 (Morgan Stanley), Gartner, Workday, Salesforce, ServiceNow, HubSpot, 汇丰银行 (HSBC), Claude Code, Codex, Vercel, Lovable, 德勤 (Deloitte) ## Video description Ashutosh Garg - CEO, Eightfold AI For twenty years, buying enterprise software meant filing requirements, waiting on a vendor's roadmap, and living with a workflow no one on the team actually chose. That era is ending: the barriers to building custom software are collapsing right alongside AI's own advances, and even experienced engineers now build with AI as their default toolkit, not the exception. The next wave of enterprise applications will come from the people who understand the problem best, armed with real domain intelligence instead of a generic model. Eightfold's TalentForge — a platform that lets teams build custom talent applications on top of deep, purpose-built domain intelligence rather than a generic LLM — brings this to life. It's a different role for a vendor to play: instead of selling you the finished app, we're handing you the domain expertise and the tools to build the one your team actually needs. Attendees will leave with a sharper framework for rethinking build versus buy, and a clear-eyed look at what it actually takes to turn domain expertise into working software. For anyone setting technology strategy for their organization, this is a shift worth getting ahead of before competitors do. [Source](https://www.youtube.com/watch?v=e-AzzfpT804)

Source quotes

[0:04] Ashutosh Garg: 重点不在于他们过去做过什么，而在于他们下一步能做什么 [2:08] Ashutosh Garg: 过去的整个企业级软件生态都建立在‘开发成本套利’的基础之上 [4:00] Ashutosh Garg: 你们有多少人真心喜欢逛街买衣服？又有谁喜欢站在试衣间外面排队试衣服？ [4:00] Ashutosh Garg: 每一次你觉得软件不顺手，就必须提交变更需求，然后陷入等待一到两年的供应商路线图排期中 [5:50] Ashutosh Garg: HR 领域的 Claude Code [12:30] Ashutosh Garg: 他就是一直在说话、说话，然后打电话继续说话，接着去开会说话，回到家又拿起电话一直说个不停 [12:30] Ashutosh Garg: 管理众多 AI Agent 的 CEO [16:49] Ashutosh Garg: 人类从来没有标准的小号、中号或大号身材，企业同样没有标准尺码

Source quotes

Ashutosh Garg: For twenty years, buying enterprise software meant filing requirements, waiting on a vendor's roadmap, and living with a workflow no one on the team actually chose. Ashutosh Garg: The next wave of enterprise applications will come from the people who understand the problem best, armed with real domain intelligence instead of a generic model. Ashutosh Garg: instead of selling you the finished app, we're handing you the domain expertise and the tools to build the one your team actually needs. [0:04] Ashutosh Garg: Looking for people who have already done the exact work is practically impossible in rapidly evolving areas like enterprise AI. Success hinges on identifying individuals who possess the foundational capabilities to learn and execute quickly. [2:08] Ashutosh Garg: When the marginal cost of creating software drops toward zero, the economic rationale for purchasing rigid, shrink-wrapped applications falls apart. [4:00] Ashutosh Garg: almost nobody enjoys waiting outside fitting rooms trying on off-the-rack garments sold in rigid sizes like Small, Medium, or Large. Human beings come in every shape and size, leaving people with closets full of clothes that do not quite fit and must be uncomfortably stretched. Enterprise software has forced companies into the exact same compromise for decades. [5:50] Ashutosh Garg: While general-purpose development assistants like Claude Code, OpenAI Codex, Vercel, or Lovable can generate code, they lack the specialized foundation necessary to operate in the talent domain. [12:30] Ashutosh Garg: He talks and talks, makes a phone call, goes to a meeting and talks, and comes home and makes another phone call. [12:30] Ashutosh Garg: In the AI era, every employee is evolving to become the 'CEO of agents,' orchestrating personal workflows and requiring custom software to execute them effectively. [16:49] Ashutosh Garg: Standard appraisal systems are backward-looking mechanisms built to penalize employees and justify compensation decisions rather than foster development.

Original transcript

[0:04] So I will talk a lot about like how we see the world changing how eight-fold is playing [0:09] a role over there. But before I dive in just to give a quick introduction of who we are, [0:14] who I am, what we are doing. The simplest way for me to talk about is 8-fold is a company [0:21] trusted by all the logos over here. some of the largest companies whether it is the [0:26] vodafone of the world to city bank of the world to the some of the ones that are not [0:31] mentioned over here whether it is Microsoft and media Netflix of the world to JB Morgan [0:35] Chase Morgan stands each of these companies are using eight-fold wall to wall either to [0:43] hire the right talent to de develop the skills of their current employees to resource the projects [0:51] to enable intermobility to career pathing to skill-based organiz ation or everything [1:00] almost like 30 to 40% of the fortune companies are relying on 8-fold top leaders in every vertical [1:08] tree and the thing that has made that possible the thing that has enabled it is a deep learning [1:13] AI technology that 8fold has pioneered over the years how we have gone ahead and started thinking [1:20] about you should not be hiring talent just based on who they are, where they have worked, [1:26] what degrees they have, but really thinking around the potential of each and every person. Not what [1:32] they have done, but what they can do next. And especially as you're walking the floors [1:37] of this hotel looking at what you can do with deep learning in your own company, in your own [1:42] organization, if you keep looking for the people who have done the work, who know AI really, really [1:47] well, it's almost impossible. So what you have to really rely on the people who can learn and [1:54] do the work quickly. So that has been the found of eightfold. We started little bit before this even [2:00] AI conference started. We believed in AI. I have been doing this for last 30 years of my career. [2:08] But the transformation we are going through in the world is very very different this time. The pace [2:13] of the change that all of us are witnessing is just very different. And the reason what [2:18] makes this makes this different this time is it's no longer around what we are building but [2:24] more foundationally how we are building things. Gartner recently published that 70% of the code [2:33] written by various vendors that you're talking over here or the enterprises is either vcoded [2:39] or written by AI or using AI agents. And what does that mean? Does that mean that we are going to go [2:48] through SAS apocalypse? Will software cease to exist? And frankly, if you were to look at the [2:56] stock price of any of your favorite companies in the SAS space, be it workday, Salesforce, [3:00] service now, actually the world is falling apart. A company like Harvest Spot is down 70%. [3:07] So what does it mean for the world of software and what has changed is the entire software [3:14] industry was built on the premise of cost of development arbitrage. It's too hard to build [3:22] software yourself too expensive. So we are better off buying it from someone else who can amortize [3:28] the cost across multiple buyers and it will be cheaper. But if the cost of building the software [3:35] has dropped, it no longer makes sense to just buy a prepackaged shink wrap software anymore. [3:44] But unlike the Wall Street, I am really in the world where I believe that we are going to go [3:49] through the golden age of software. And to better understand why I believe that, [3:55] why I think we are going to enter that phase, I know it's a small audience, [4:00] but I will still ask everyone, right? How many of you love shopping? [4:06] And even better, right? How many of you love standing outside the fitting [4:11] room and then trying out the clothes? No, you don't. Anyone who loves that? [4:21] Maybe there's one. And the reason is see when you buy like when I go [4:26] and buy chauffeer shirts or dress I mean trousers right it's a small medium large. [4:34] But I wish God has created us in three sizes. A small medium large. When my [4:40] kids are growing I'm like let me choose a medium for him and large for this one [4:44] right? No we come in all shapes and sizes. And then we the result is we [4:50] have a wardrobe full of clothes that don't fit us that kind of look good and we are [4:55] stretch them little bit here and there. Right now think of your own organization. [5:03] Is City Bank City Bank because they do everything exactly like HSBC? No. [5:08] Do they have the same culture, same people, same technology stack? No. [5:14] But then you are forced to buy the same CRM, same HRIS, same financial planning software and spend [5:24] years configuring it to fit your needs and then every time you feel like it's not working for you, [5:32] you file a request, change request, which might take a year, two year road map. That is a cycle [5:39] all of us go through and I think that needs to change and that can change now and that is what [5:50] at 84 we are doing is foundationally changing that entire concept of SAS application software [5:56] industry by introducing the concept of talent fold a software that enables you to build applications [6:06] for yourself enables you to build applications customtailored for you for your organization. [6:17] I mean I don't like comparisons but in this again audience right simplest [6:22] way to think about what we do is can you have claude code for HR? [6:31] No. [6:34] So now what why now the question might be why do you even need cloud code for HR why [6:42] can't you just use cloud code or codeex or war cell or loveable or favorite application that [6:47] you have right because one the talent agents the understanding that we have collected from billions [6:56] of data points billions of career paths over the year to understand who has performed well who is [7:01] likely to fit who is likely to be able to learn the skills quickly becomes a foundation of this [7:07] entire platform. So now you no longer have to rely on keywords but you can really see through [7:14] people think through what they can do how and where they can succeed. And the second thing [7:20] is the trust. Trust is the core one of the core themes today in the hallways in the expo hall. [7:27] Right? But what do we mean by trust? It's not only about the model trust. It's not only about [7:35] the observability of the model. I mean that is extremely important. But equally important, [7:41] if not more important, is the trust in the data that you'll use and the trust in the workflows [7:47] and the actions the agents that you're building on top of it can make. Here is a very simple [7:53] example. Especially in the HR space, if you're building on the recruiting side, you may say, [7:58] "Oh, you know what? I'm going to custom build it. And one of the things that I will do is I [8:03] will use age of applicant. People have done that. No, you cannot do that. Okay, fine. I will not do [8:14] that. But what I will do is as the candidates are going through an interview process, if everyone [8:20] is a thumbs down, I'll just go and send them an autorejection email. No, you cannot even do that. [8:29] or in the interview process I have my agent ask for the home address and family members [8:33] and everything about the candidate. No, you cannot do that. So having that trust inbuilt into [8:42] the platform especially in HR where compliance matters, governance matters, individuals career [8:47] paths matters becomes extremely central. But if you do that you can make magic happen and that [8:57] is what we enable with talent forge and as you see the next few slides how as I show you what [9:05] we are doing I won't be surprised if most of you are work your the companies that you work for are [9:13] already using eightfold if you work for one of the trillion dollar companies almost 70% likelihood [9:21] you your company is using 8fold and you have already access to it. If you are one of the top [9:27] banks, yes, you have access to 8fold. If you're one of the top hospitals or pharma companies, [9:33] you have access to 8fold. So what you can do is when you go back, you don't have to rely [9:40] on the HR software that has been provided to you. You can build if you're a manager, [9:45] you can build application for yourself, for your team. If you're an employee, you can build it for [9:50] yourself. If you're a HR provider, you can build it. So very simple how it works. It's a developer [10:01] dashboard. You can come in in simple language. You can describe what you want. So very first thing is [10:08] just describe. I want a performance management system. I want an analytics dashboard. I want [10:13] a workflow as a manager to manage and understand the skills and the development plans of my team. [10:21] Whatever application you want to build, you can just describe it. [10:29] You don't need to know any coding. Now system will further help you think through [10:35] what you have described is correct. Just get a bit more details from you. [10:46] And if you're wondering is this for me as I'm [clears throat] describing it uh one [10:51] of our junior HR employee ended up building the entire HRIS by themselves in a matter of a week. [11:04] Now this is the most important one. Based on what you describe, can it automatically come up with a [11:10] functional requirements to make that happen? And not just a functional requirements are [11:15] nonfunctional requirements around compliance, scalability, governance, trust, bias, ethics, [11:25] everything matters in HR because ultimately what you are building might impact people's career. [11:32] It may impact their pay, performance, hiring, any of those things. So what it will do is it [11:43] will identify what are the challenges you might run into in your requirements. Are we focused on [11:48] that or not? How you should modify those things. Then that will help you visualize what you have [11:56] built and if you like it otherwise you can go back modify the requirements. It will take [12:04] you through the whole process of deploying it, testing it and publishing it. A simple [12:11] application can be built by you in a matter of hour. You no longer have to rely on anyone else [12:25] and you just publish. [12:30] Now why this is actually very foundational? [12:37] Why I as an employee shouldn't have my own personal software? Why do I have [12:44] to rely on a prepackaged software that I have got? [12:50] Everyone has a different strength, different weaknesses. [12:54] And in fact, if you really think about it, where the world is headed to put it in perspective, [13:00] I'll give you another analogy and example. [13:04] If you I have two boys 9 and 12 year olds. If you go and ask them what does daddy do? [13:14] I wish they say that okay you know what he's running a company and he runs go to market [13:18] and manage the processes and comes up with the product strategy and plan and sales and [13:22] everything but no they won't say that. What they will say is he talks and talks and then [13:29] makes a phone call and talks more and then goes to a meeting and talks and then comes back home [13:34] and again makes a phone call and talks and talks and talks and talks which actually is quite true. [13:44] And once you take that perspective what we are doing as human beings [13:48] as we go up in our career right we end up working through people [13:55] And the role of each and every one of us is evolving in to be the CEO of agents. [14:03] Everyone is going to just manage that and for that we will need our own software, [14:09] our own infrastructure that can enable that to do [14:12] thing in the way we want it. Right? So here's an example. Onboarding [14:21] a typical company onboarding time takes almost three months for a new hire to be effective. [14:31] Not just that every department the way you want to onboard a software engineer is very [14:35] different the way you want to onboard a finance analyst and a very different [14:40] than a HR recruiter or someone else. But then the way you want to onboard the same [14:46] software engineer in Mountain View is very different than the way you want to onboard [14:49] that engineer in let's say London. But then it's also very different than if you want to [14:55] onboard someone who has a 15 years of experience and someone who is a fresh college graduate. [15:02] So now you can build a personalized [15:05] onboarding experience for each individual each team in no time. [15:15] And what is interesting over here is there are companies that have spent [15:20] decades just building an onboarding solution and selling to it. Right? [15:24] But now all that can be coded through by talent forge in no time by you. [15:32] This is an interesting one. Performance review. [15:36] Is there anyone in this room who loves performance review? Anyone [15:43] who likes to receive performance review, [15:44] let's come and talk about the performance. No, actually even worse, [15:52] it's so hard to have a tough conversation, right? No one likes enjoys that either. [15:59] And the reason is these solutions are just so broken. They're designed for a different era. They [16:06] are backward-looking solutions. They are about to penalize. They're about to comp. What people [16:13] want is career development. Let's talk about how I can coach and mentor. How can I help you grow? [16:23] The way I want to performance manage or have this conversation with an account executive is [16:27] very different from a software engineer than a product manager than a marketing analyst versus [16:31] content writer. Right? So why can't I have a custom-designed software for each employee [16:39] in my own company and actually on the previous one uh as you are talking about it enterprises [16:49] like telecom one of the largest telos in the world are building this with talent forge deoid [16:57] one of the largest services company in the world is building it with talent forge and so [17:05] Anyone in the measures in this room, if you can get a view of your team the way [17:10] you want it every Monday morning so that you can be more effective in guiding them [17:15] and developing them in delivering the outcomes, minimizing the surprises. [17:25] Even the simplest thing is if I can know in my deck reports who is doing one-on- ones and who [17:30] is not it will just change the performance of the whole team but the only option today [17:37] I have without okay fine we need to put out RFP we have to spend two years buying a software we [17:44] will have to shell out half a million dollars and then it will take us another two years to [17:47] implement and that whole world is changing and that is why I believe that we are genuinely in [17:53] the age golden age of software with we can custom build the thing the way it should be [18:01] because neither the humans come in small, medium and large sizes nor our enterprises [18:06] come in those sizes. All of us are unique. That has what made this world colorful, exciting, [18:12] interesting place to thrive and that is why each of your organizations are thriving and [18:17] that is why you need this thing. So the future is going to be built by you and to [18:24] see it in action come to booth 1354. If you are taking time off today we will be there tomorrow. [18:31] Please come visit us. Would love to show you what we are doing. With that thank you all.