Autodesk Cuts Procurement Question Time 80% via Direct Code Queries
Ai4
Autodesk measures and scales customer trust by publishing nutrition-label transparency cards for AI features and querying product codebases directly to accelerate security reviews.
## Background
This content was published on YouTube by Ai4 on August 21, 2026. The page description outlines a fireside chat between Navrina Singh (CEO/Founder of Credo AI) and Kate Adam (Director of Customer Trust & Employee Awareness at Autodesk) exploring how organizations can pair AI innovation with robust governance, transparency, and employee awareness.
## TL;DR
In this discussion, Navrina Singh and Kate Adam examine how enterprise AI adoption hinges on building measurable trust rather than just chasing technical capability. Adam outlines how Autodesk tracks customer trust via four key factors (Humanity, Transparency, Capability, Reliability) and uses standardized 'Transparency Cards' modeled after nutrition labels to clarify AI features and data usage. She details Autodesk's internal shift toward becoming AI-native through rapid agentic experimentation—such as querying software codebases directly to complete procurement questionnaires—and emphasizes holistic governance spanning internal tools, security, and customer transparency.
## Atmosphere & style
The presentation is conversational, engaging, and highly energetic, blending lighthearted moments with deep, practical insights. The rapport between Singh and Adam is warm and collegial, opening with playful banter about Swiftie culture and custom friendship bracelets before transitioning into an authoritative discussion on security and enterprise governance. The tone is practical, forward-looking, and candid, creating an inviting yet professional environment that encourages fast experimentation, industry collaboration, and authentic transparency.
## Key takeaways
- [00:04](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=4s) Autodesk establishes trust through a bridge of internal disciplines—such as Infosec, Privacy, Responsible AI, and GRC—while maintaining a direct customer feedback loop via its public Trust Center.
- [00:04](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=4s) A primary obstacle preventing enterprises from fully deploying AI or becoming AI-native is the 'trust overhang' rather than technical capability.
- [04:33](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=273s) Autodesk measures customer trust using a framework derived from Deloitte's 'Four Factors of Trust': Humanity, Transparency, Capability, and Reliability.
- [08:20](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=500s) Autodesk uses standardized 'Transparency Cards'—formatted like nutrition labels—on its Trust Center to clearly display training datasets, encryption, guardrails, and inference details for AI features.
- [11:27](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=687s) Becoming 'AI-native' requires shifting from traditional Agile to an experimental mindset where teams ship incremental improvements and embrace fast failures.
- [11:27](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=687s) By shifting to an agentic 'ask the product' workflow that queries codebase data directly, Autodesk reduced customer security questionnaire processing time by 5x.
- [16:38](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=998s) Autodesk tracks internal and external alignment using a single key survey metric: 'How clearly do you understand what Autodesk is doing with your data and why?'
- [16:38](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=998s) Organizations should avoid operating in silos when managing AI governance and actively share failure and success stories across the industry.
- [16:38](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=998s) Enterprise AI trust must extend beyond customer trust to encompass internal AI tools, development practices, and governed product features.
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=4s) Introduction and Taylor Swift Friendship Bracelets
Navrina Singh, founder and CEO of Credo AI, opens the session by welcoming Kate Adam, Director of Customer Trust & Employee Awareness at Autodesk. Navrina emphasizes that Autodesk is proving how trust acts as a growth driver and how AI governance serves as a launchpad for innovation. She playfully reveals that both she and Kate are massive "Swifties," noting that Kate brought custom "trust" friendship bracelets to pass out to attendees after the session. Kate quips that she couldn't stop making friendship bracelets after the Eras Tour ended, leading her to channel that energy into Autodesk trust friendship bracelets to keep the habit productive.
Turning to serious matters, Navrina highlights that getting frontier-class technical capability is relatively easy today, but the true hurdle is the "trust overhang." This hesitation prevents enterprises from fully deploying AI or becoming AI-native. Kate explains that her role as Director of Customer Trust was established because Autodesk develops cloud-based products and 3D geometric AI models for the design and make industry. Customers who entrust Autodesk with terabytes of highly confidential data—such as architectural blueprints, complex manufacturing processes, and major film designs—need absolute assurance that their data and confidentiality are protected.
- **Internal Trust Disciplines:** Kate’s team acts as a bridge uniting internal divisions, including Information Security (Infosec), Application Security, Privacy, Responsible AI, Resilience, Governance, Risk and Compliance (GRC), and Trust Architecture.
- **Customer Feedback Loop:** Kate’s group runs Autodesk’s public Trust Center and speaks directly with customers ranging from Fortune 100 corporations to "mom-and-pop" shops. They collect security and governance use cases to feed directly back to internal product development teams.
[04:33](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=273s) Defining and Measuring Trust via the Four Factors
After polling the audience to see how many have heard the phrase "AI trust" in the past 24 hours, Navrina poses a core question: since Autodesk customers design buildings, aircraft, and infrastructure built to last 30+ years, what does *measurable* trust look like? Kate responds that trust is quantifiable and directly impacts customer loyalty, renewals, and revenue. Autodesk tracks trust by conducting quarterly surveys across all customer segments using a framework derived from Deloitte research published in the book *The Four Factors of Trust*.
Kate details how they track four primary metrics to evaluate customer sentiment and identify exact focus areas. Her team pays special attention to transparency, ensuring customers understand how their data is secured in cloud products, how AI models function, and where safeguards exist. Navrina reinforces this by noting that over six years of building the AI governance category, Credo AI has seen that companies investing in the infrastructure of trust are the ones that ultimately win.
- **Humanity (Caring):** Gauging whether customers feel Autodesk cares about them beyond their next purchase order.
- **Transparency:** Ensuring customers know exactly what Autodesk is doing with their data and how processes work.
- **Capability:** Assessing whether customers believe Autodesk products and teams can deliver on their promises.
- **Reliability:** Tracking whether Autodesk consistently does what it promises to do, when it promises to do it.
[08:20](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=500s) Transparency Cards: AI Nutrition Labels
Navrina highlights Autodesk’s "Transparency Cards" as a prime example of an artifact that turns governance into concrete, measurable trust. Kate explains that as Autodesk accelerated its internal adoption and external development of AI, internal vendor teams and external customers alike demanded transparency rather than black-box models. To address questions about training datasets, encryption, guardrails, inference locations, and specific features, Autodesk rejected publishing dense blocks of web text across its massive product catalog. Instead, they adopted the ubiquitous format of nutrition labels.
These transparency cards are hosted on Autodesk's Trust Center for every AI feature, instantly informing users of essential "nutrition facts" in a standardized, recognizable format. Kate emphasizes that these cards are not static documents; they evolve continuously as customer and market signals demand deeper insights. This approach provides transparency without heavy administrative strain and has trained internal engineering teams to expect that every new AI feature requires its own transparency card.
[11:27](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=687s) Becoming AI-Native and the "Ask the Product" Experiment
Addressing the session's overarching theme of building trust in enterprise AI, Navrina asks what being "AI-native" means inside Autodesk, particularly when rigorous transparency collides with engineering demands. Kate shares insights from a recent leadership offsite, defining "AI-native" not as using AI for everything—much like you wouldn't use a hammer for every construction task—but as recognizing AI as a powerful tool for appropriate use cases. Culturally, becoming AI-native requires shifting from traditional Agile to a mindset where teams ship with incremental 20% to 30% improvements, embracing fast experimentation and quick failures over slow perfectionism.
Kate illustrates this shift with a real-world internal success story regarding customer security questionnaires during procurement. Previously, answering complex questionnaires across Autodesk's extensive product portfolio required manual outreach to engineers, taking weeks to process. Kate’s team hypothesized an agentic workflow: rather than asking human engineers, they could "ask the product" and query the codebase directly.
- **Procurement Efficiency:** Four months into the experiment, the team completes customer security questionnaires with higher accuracy and a 5x reduction in time—all without disturbing engineering teams.
- **Expanded Use Cases:** Querying the codebase directly has expanded into automatically generating white papers, mapping software dependencies, and analyzing technical components.
- **Speed of Innovation:** Failing fast within days is vastly preferable to spending two months trying to reach perfection, especially in a fast-moving ecosystem where standards like Model Context Protocol (MCP) emerged only within the past year.
[16:38](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=998s) Rapid-Fire Q&A: Governance Metrics, Industry Silos, and Open Source
Navrina transitions to a rapid-fire round to extract actionable takeaways in the session's final minutes. She prompts Kate with short, direct questions regarding executive reporting, common pitfalls in AI governance, open-source model trust, and a final definition of enterprise trust.
- **The One Transparency Metric:** Kate shares the exact survey question she displays on every slide for privacy lead, AI builder, data science, and executive alignment: *"How clearly do you understand what Autodesk is doing with your data and why?"*
- **What NOT to Do in AI Governance:** Do not operate in silos or assume you are hopelessly behind. Kate urges organizations to share stories of failure and success across the industry—even with competitors—because most enterprises are navigating the AI maturity curve together.
- **Foundation Models vs. Open Source:** When asked which she trusts more, Kate bluntly answers, "I'm a security person at heart... I don't trust open source," prompting a laugh and an offer from Navrina to debate the topic further.
- **Enterprise AI Trust Is Not...:** Kate completes the sentence by asserting that enterprise AI trust is not *only* customer trust; it encompasses internal AI tool usage, internal development, product features, and practical use cases governed holistically so organizations can monitor and pivot quickly.
## Notable quotes
> [16:38](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=998s) How clearly do you understand what Autodesk is doing with your data and why? — Kate Adam
> [16:38](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=998s) I'm a security person at heart... I don't trust open source — Kate Adam
## Mentioned
Navrina Singh, Kate Adam, Autodesk, Credo AI, Taylor Swift, Eras Tour, Deloitte, The Four Factors of Trust, Trust Center, Transparency Cards, Model Context Protocol
## Video description
Kate Adam - Director, Customer Trust & Employee Awareness, Autodesk
Navrina Singh - CEO/Founder, Credo AI
As AI moves from experimentation to enterprise-wide adoption, trust cannot be treated as an afterthought. This fireside conversation between Navrina Singh, Founder and CEO of Credo AI, and Kate Adams, Director of Customer Trust & Employee Awareness at Autodesk, will explore how organizations can connect AI innovation with clear governance, transparency, and employee awareness. The discussion will examine both the human and operational dimensions of building confidence in AI, including how leaders can create shared accountability across teams. Attendees will leave with practical considerations for scaling AI responsibly while maintaining trust with employees, customers, and other stakeholders.
Autodesk通过直接查询代码将采购问答耗时缩短80%
Autodesk通过发布针对AI功能的“营养标签”式透明度卡片，并直接查询产品代码库以加速安全审查，以此衡量并拓展客户信任。
在本次讨论中，Navrina Singh与Kate Adam探讨了企业落地AI的关键在于建立可衡量的信任，而非单纯追求技术能力。Adam概述了Autodesk如何通过四个关键要素（人性、透明度、能力、可靠性）追踪客户信任，并采用效仿营养标签的标准化“透明度卡片”来说明AI功能与数据使用情况。她详细阐述了Autodesk如何通过快速的智能体实验向AI原生转型——例如直接查询软件代码库来填写采购问卷——并强调了涵盖内部工具、安全及客户透明度的全面治理。
Credo AI 与 Autodesk 指出，阻碍企业全面部署 AI 或成为 AI 原生企业的主要障碍并非技术能力，而是“信任悬崖”（trust overhang）。 (0:04)
Autodesk 借鉴德勤的“信任四要素”（人性关怀、透明度、能力、可靠性）框架按季度量化客户信任，并在其公开信任中心通过类似营养成分表的标准化“透明度卡”（Transparency Cards）展示 AI 特性。 (4:33)
在向 AI 原生转型的过程中，Autodesk 推行了基于智能体的“直接询问产品”（ask the product）工作流，通过直接查询代码库将客户安全调查问卷的处理时间缩短了 5 倍。 (11:27)
Autodesk 通过统一的关键调研指标“你有多清楚了解 Autodesk 如何处理你的数据以及原因？”来对齐内外部沟通，并呼吁企业在 AI 治理中打破孤岛、积极分享行业经验。 (16:38)
Reading summary
Enterprises face a significant trust overhang that inhibits full AI adoption, requiring organizations to treat governance and transparency as quantifiable growth drivers rather than administrative burdens, according to Autodesk's Kate Adam and Credo AI's Navrina Singh.
Measuring Enterprise Trust Through Four Quantifiable Factors (4:33)
Because Autodesk manages sensitive design, manufacturing, and architectural data intended for assets that endure for decades, customer assurance directly influences renewal rates and revenue. Adam explains that Autodesk measures this dynamic through quarterly customer surveys adapted from Deloitte's framework in 'The Four Factors of Trust,' which tracks Humanity, Transparency, Capability, and Reliability.
Within this framework, Adam prioritizes transparency regarding how cloud products protect customer information and deploy AI models. To evaluate cross-functional alignment across executives, data scientists, and privacy teams, Autodesk tracks customer sentiment against a single core benchmark: whether customers clearly understand what Autodesk is doing with their data and why.
Standardizing AI Disclosures with Transparency Cards (8:20)
To satisfy customer and vendor demands for visibility into AI functionality without relying on opaque models or dense text documentation, Autodesk implemented standardized 'Transparency Cards' on its public Trust Center. Designed around the concept of nutritional facts labels, these cards present essential operational details across every AI feature in Autodesk's catalog.
Each card details training datasets, guardrails, encryption methods, and inference locations. Adam notes that these artifacts are living disclosures that update according to ongoing customer feedback, while internally conditioning software engineering teams to routinely design transparency documentation into every new AI feature release.
Accelerating Internal Workflows via Codebase Querying (11:27)
Adam defines becoming an AI-native organization as selecting AI specifically for appropriate tasks and abandoning slow perfectionism in favor of shipping 20% to 30% incremental improvements. Rather than taking weeks to manually interview engineers to complete complex customer security questionnaires during procurement, Autodesk tested an agentic workflow titled 'Ask the Product.'
Four months into implementation, querying the codebase directly reduced procurement questionnaire completion time by 5x while increasing answer accuracy and preventing engineering interruptions. Autodesk has since expanded this direct codebase interrogation to map software dependencies, inspect technical components, and automatically generate white papers, prioritizing fast, days-long failure cycles over months of planning.
Navigating Governance Pitfalls, Scope, and Model Security (16:38)
Addressing common governance mistakes, Adam cautions against operating in corporate silos or assuming peers are significantly further ahead on the AI maturity curve, urging organizations to openly share both operational successes and failures across the industry. When asked about underlying infrastructure, Adam stated that her security background leads her to distrust open-source models in comparison to foundation models, a perspective Singh noted remains an active industry debate.
Adam concludes that enterprise AI trust must not be confined to outward customer perception. Instead, it requires holistic internal governance covering employee tool adoption, in-house software development, and active feature monitoring so organizations can rapidly detect issues and pivot strategy.
While Autodesk demonstrates measurable efficiency and procurement gains by querying code directly and standardizing transparency disclosures, unresolved questions remain regarding model selection, specifically balancing open-source models against proprietary foundation models from an enterprise security standpoint.
精炼阅读摘要
Credo AI 创始人兼首席执行官 Navrina Singh 与 Autodesk 客户信任总监 Kate Adam 指出，企业规模化应用人工智能的核心瓶颈是“信任滞后”而非前沿技术能力。Autodesk 通过量化信任指标、构建标准化的透明度卡片，并推行敏捷的内部智能体治理，旨在将 AI 信任转化为驱动企业业务增长与创新的基础设施。
化解信任滞后并整合跨职能治理机制 (0:04)
Navrina Singh 指出，尽管当前获取前沿级技术能力的门槛相对较低，但企业在全面部署 AI 或转型为 AI 原生企业时，普遍面临着“信任滞后”（trust overhang）的顾虑。Kate Adam 解释称，Autodesk 开发面向设计与制造行业的云原生产品及 3D 几何 AI 模型，其客户群体（涵盖建筑、精密制造与影视制作领域）托付了大量涉及机密蓝图和复杂流程的数据资产，因而对数据隔离与机密性保障提出了极高的合规要求。为此，Autodesk 设立了专门的客户信任团队以应对安全挑战。
采用信任四要素框架建立量化评估体系 (4:33)
针对建筑、飞行器等需维系 30 年以上工程周期的设计场景，Autodesk 认为信任并非抽象概念，而是直接关联客户忠诚度、续约率和营收的可量化指标。公司基于德勤在《信任的四大要素》（The Four Factors of Trust）一书中提出的理论框架，面向各层级客户开展季度调研，重点追踪“人性关怀”（评估客户是否感受到被重视）、“透明度”（客户是否清楚数据用途）、“专业能力”（产品能否兑现承诺）与“可靠性”（交付的稳定度与时效性）。
推出透明度卡片建立模型营养标签规范 (8:20)
为消除企业客户与内部团队对黑盒模型的疑虑，Autodesk 摒弃了在庞大产品目录中堆叠晦涩文本的做法，借鉴食品“营养成分标签”形式推出了标准化的“透明度卡片”（Transparency Cards）。这些卡片公开发布在 Autodesk 信任中心，为每项 AI 功能清晰列明训练数据集、加密措施、安全护栏设置、推理发生地点及具体功能参数。
培育 AI 原生文化并借代码库问答提效 (11:27)
在探讨企业如何成为“AI 原生”时，Kate Adam 表示，AI 原生并不意味着将 AI 滥用于所有场景，而是将其视为特定任务的高效工具。在组织文化层面，这要求团队突破传统敏捷开发对长期打磨的依赖，转向追求实现 20% 至 30% 的渐进式改进，接受快速试错而非追求完美。例如，Autodesk 针对采购阶段耗时费力的安全调查问卷，启动了“向产品提问”（Ask the Product）的智能体工作流实验。
打破治理孤岛并确立全方位企业信任标准 (16:38)
在总结 AI 治理实践时，Kate Adam 告诫企业切勿闭门造车或因焦虑而自乱阵脚，呼吁行业同行乃至竞争对手之间积极交流成功与失败经验，因为各家机构目前都在共同摸索 AI 成熟度曲线。就模型选型倾向而言，Kate Adam 基于安全专业背景明确表达了对开源模型安全性的谨慎与不信任，认为基础模型在企业级受控环境中更具把控力。
Kate Adam 总结指出，企业级 AI 信任不能仅仅窄化为客户维度的信任，而必须全面覆盖内部工具采购、代码自研与对外产品交付的全生命周期。对于开源模型在企业复杂环境中的可靠性与风险控制，安全领域与开发领域仍存在尚未完全弥合的审慎分歧。
Detailed notes — zh
## Background
该内容由 Ai4 于 2026 年 8 月 21 日发布在 YouTube 平台上。根据页面简介，该内容记录了 Credo AI 创始人兼 CEO Navrina Singh 与 Autodesk 客户信任与员工意识总监 Kate Adams 的对谈，主要探讨企业如何将 AI 创新与明确的治理、透明度及员工意识相结合，打造覆盖全流程的企业级信任体系。
## TL;DR
本次对话围绕企业从 AI 实验走向大规模应用过程中面临的“信任悬崖”展开。Autodesk 分享了其如何通过德勤“信任四大要素”框架量化客户信任，并创新性地推出类似营养成分表的“AI 透明度卡片”，以直观公开模型训练与数据使用细节。此外，探讨了在敏捷开发向 AI 原生转型中采用智能体工作流（Agentic Workflow）提升合规效率的实践，强调 AI 信任并非事后补充，而是推动业务增长与产品创新的核心保障。
## Atmosphere & style
整场对话氛围轻松而专业，以两位嘉宾同为泰勒·斯威夫特粉丝并制作“信任友谊手环”的幽默破冰拉开序幕。随后的探讨兼具行业前瞻性与接地气的实操干货，充满着同频共振的技术共鸣与深度洞察。结尾的高密度快问快答则将气氛推向高潮，节奏明快、言简意赅，为观众带来一场既亲切又极具说服力的企业治理经验分享。
## Key takeaways
- [00:04](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=4s) Autodesk 客户信任与员工意识总监 Kate Adams 与 Credo AI 创始人兼 CEO Navrina Singh 探讨了企业在 AI 规模化应用过程中建立信任与治理体系的重要性。
- [00:04](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=4s) 企业在获取前沿 AI 能力时遇到的瓶颈通常不是技术本身，而是“信任悬崖”（Trust Overhang），即对安全和信任的顾虑阻碍了企业部署与转型。
- [00:42](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=42s) Autodesk 设立“客户信任总监”作为内部信任安全团队与外部客户之间的桥梁，将客户的治理与安全需求转化为产品功能与反馈闭环。
- [00:42](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=42s) 信任并非虚无缥缈，Autodesk 引入德勤“信任四大要素”（人性关怀、透明度、业务能力、可靠性）框架，通过季度调查量化客户信任得分。
- [08:25](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=505s) 为提高透明度并解决“黑盒”担忧，Autodesk 推出了借鉴食品“营养成分表”形式的“ transparency cards”（透明度卡片），公示 AI 模型、数据集、防护措施及训练推理位置等信息。
- [11:27](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=687s) 成为 AI 原生企业并不意味着盲目套用 AI，而是转变领导者思维与运营模式，接受 20%-30% 的渐进式创新，通过敏捷试验快速学习与迭代。
- [11:27](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=687s) Autodesk 团队利用智能体工作流（Agentic Workflow）自动化处理繁琐的客户安全问卷，在 4 个月内将处理速度提升了 5 倍。
- [16:40](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=1000s) 企业在 AI 治理中应避免陷入盲目焦虑，大多数企业都在摸索阶段，行业间与竞争对手间应多交流分享实践经验。
- [16:40](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=1000s) Kate Adams 认为 AI 信任不仅仅关乎客户信任，它就是业务本身，涵盖内部工具、研发、产品集成和监控治理的全流程。
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=4s) 泰勒·斯威夫特友谊手环与企业级 AI 信任的悬崖
会议在轻松诙谐的氛围中拉开帷幕。Credo AI 的创始人兼 CEO Navrina Singh 介绍了本场对话的嘉宾——Autodesk 客户信任与员工意识总监 Kate Adams。Navrina 称赞 Autodesk 正在用实际行动证明“信任是增长的驱动力，而 AI 治理是 AI 创新的发射台”。随后，她透露了一个轻松的幕后细节：两位都是泰勒·斯威夫特（Taylor Swift）的忠实粉丝（Swifty）。在“时代巡回演唱会”（Eras Tour）结束后，Kate 甚至在公司内部掀起了一股制作“Autodesk 信任友谊手环”的狂热潮，Navrina 特别提示现场听众，会后可以找 Kate 索取定制手环。
在轻松的破冰过后，话锋转向了当前企业级 AI 所面临的核心阵痛。Navrina 指出， Kate 拥有当下 AI 领域“最艰难的工作”之一。因为对于企业而言，获取最前沿的 AI 能力其实并不困难，真正沉重的瓶颈在于“信任悬崖”（Trust Overhang）。这种对安全与信任的顾虑，不仅阻碍了企业部署与采用前沿 AI，更成为了企业向 AI 原生化（AI Native）转型的主要障碍。
- **Credo AI**：AI 治理与信任平台的领先提供商。
- **Autodesk**：设计与制造软件巨头，正积极将其 3D 几何 AI 模型及云端产品与 AI 治理相融合。
---
### [00:42](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=42s) 连接内部安全与外部客户的“信任桥梁”
作为 Autodesk 客户信任总监，Kate 详细阐述了这一特殊岗位的职责与诞生背景。Autodesk 的产品深度服务于设计与制造（Design and Make）行业，涵盖三维几何模型及庞大的云端协作。客户需要将 TB 级别的敏感数据托管给 Autodesk，其中包含极其机密的建筑设计、工业制造流程乃至尚未上映的电影巨制。
为了消除客户对数据保护和保密性的担忧，“客户信任总监”这一角色应运而生，成为了内部专业团队与外部客户之间的关键桥梁。 Kate 总结道，她将客户的实际需求与情绪反哺给产品团队，打造了一个促成业务变革的反馈闭环。
```
+-----------------------------------------------------------------+
| Autodesk 内部信任安全组织 |
| (信息安全 | 应用安全 | 隐私 | 负责任 AI | 合规 | 信任架构) |
+-----------------------------------------------------------------+
│
▼
[ Kate Adams 团队 (客户信任) ]
(运营 Trust Center / 收集客户反馈闭环)
▲
│
+-----------------------------------------------------------------+
| 外部客户 (从 Fortune 100 到小微企业) |
| (关注数据安全、保密性、设计图纸与制造流程保护) |
+-----------------------------------------------------------------+
```
- **内部信任组织构成**：信息安全（Infosec）、应用安全、隐私保护、负责任 AI（Responsible AI）、业务韧性、治理风险与合规（GRC）以及信任架构（Trust Architecture）。
- **产品端赋能**：将客户的安防与治理需求直接转化为产品功能（Use cases），允许客户利用自身治理体系管理其软件栈。
---
### [00:42](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=42s) 量化信任：德勤“信任四大要素”在企业的落地
针对 Navrina 提出的“如何量化信任”这一核心问题，Kate 强调信任绝非虚无缥缈的情感，而是完全可以进行科学量化的指标。Autodesk 采用了源自德勤（Deloitte）两位专家在《信任的四大要素》（*The Four Factors of Trust*）一书中提出的理论框架，按季度向所有客户群体发放问卷调查，以评估企业在客户心中的信任得分。
Kate 解释称，客户的信任度与用户的忠诚度、续约率及企业营收直接挂钩。在四大要素中，Kate 的团队尤其关注“透明度”这一指标，通过季度数据监控及时调整策略，让 Autodesk 成为“可信赖”的代名词。
| 信任要素 (Factor) | 核心含义与评估问题 |
|:--- |:--- |
| **人性关怀 (Humanity)** | 企业是否真正关心客户，而不仅仅关注下一张采购订单（PO）。 |
| **透明度 (Transparency)** | 客户是否清晰了解企业在做什么、如何处理其数据与云端产品。 |
| **业务能力 (Capability)** | 产品的性能是否达标，企业是否有能力兑现其承诺的功能。 |
| **可靠性 (Reliability)** | 企业能否持续、稳定地“言出必行，按时交付”。 |
---
### [08:25](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=505s) AI 营养标签：“透明度卡片”的创新实践
在探讨如何提升透明度时，Kate 分享了 Autodesk 推出的“透明度卡片”（Transparency Cards）。在内部研发和外部引入 AI 的过程中，Autodesk 同样遇到了“黑盒”担忧——内部团队和外部客户都迫切想知道数据究竟是如何被使用的。为了打破这种沉闷，Autodesk 借鉴了大家熟知的食品“营养成分表”（Nutrition Labels）形式。
与其在官网上堆砌没人愿意阅读的冗长文本，Autodesk 针对每一个 AI 功能模块，制成了直观的“AI 营养卡片”并公开发布在 Trust Center 上。这些卡片并非一成不变，而是根据市场反馈动态迭代，成为了内部产品团队研发 AI 新功能时的标准动作。
- **透明度卡片公示的核心信息**：
- 使用了哪些具体的 AI 模型与数据集。
- 功能的实际用途与安全防护措施（如加密与安全护栏 Guardrails）。
- 模型训练（Training）与推理（Inference）发生的确切位置。
---
### [11:27](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=687s) 从敏捷开发到 AI 原生：允许 20%-30% 的渐进式创新
谈及企业如何成为“AI 原生”（AI Native），Kate 澄清了一个普遍误区：AI 原生并不意味着在所有地方都强行使用 AI，而是将 AI 视作工具箱中一把特定而强大的“锤子”，在恰当的场景下精准施展。更重要的是，AI 原生代表着企业运营模式与领导者思维方式的根本转变。
传统的敏捷开发（Agile）正在向 AI 原生模式演进。企业不再追求耗时数月去打造“完美”的产品，而是接受以 20% 至 30% 的渐进式改善快速发布，通过“快速失败”（Fail Fast）和频繁试验来加速创新。减速试验是为了最终的加速，打破常规只要能带来学习成果就是有价值的。
- **客户安全问卷自动化实验**：
- **背景痛点**：以往处理采购环节的客户安全问卷极度依赖人工，涉及众多复杂产品线，需要跨部门向工程师核实，耗时数周且无法规模化。
- **智能体工作流（Agentic Workflow）尝试**：团队改变思路，直接通过代码和 Agent 询问产品本身“是否具备某种安全特性”，并接受早期 20%-30% 的不完美提升。
- **实验成果**：经过 4 个月的迭代，问卷完成的准确率大幅提高，处理速度**提升了 5 倍**，且完全无需打扰底层研发工程师。
- **衍生收益**：衍生出了自动生成技术白皮书、快速精准梳理架构组件与依赖关系（Dependencies）等全新应用场景。
---
### [16:40](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=1000s) 快问快答与总结：企业 AI 信任的全貌
在对话的最后三分钟，Navrina 与 Kate 展开了一轮高密度的“快问快答”（Rapid Fire），凝练了关于治理与信任的极简见解：
- **核心展示指标**：如果要在 PPT 上向隐私、AI 开发者、数据科学及信任部门展示同一个指标，Kate 会选择让客户回答这一问题：“*你是否清晰了解 Autodesk 正在如何使用你的数据以及为什么使用？*”
- **治理避坑指南**：不要假设你的企业在 AI 成熟度曲线上极度落后。虽然有极少数领跑者，但绝大多数企业都在摸着石头过河。行业间（甚至竞争对手之间）应当打破孤岛，多分享失败与成功案例。
- **基础模型 vs 开源模型**：当被问及更信任基础模型厂商还是开源模型时， Kate 坦言“我骨子里是个安全人员，我不信任开源。”
- **金句填空**：“企业 AI 信任不仅仅是……” Kate 的回答是：“*企业 AI 信任不仅仅是客户信任，它就是业务本身。*” 它涵盖了内部工具使用、AI 研发过程、产品集成以及监控治理的全流程融合。
---
<ElicitationsGroup message="如果您想进一步了解企业 AI 治理与信任落地：">
<Elicitation label="深入分析 AI 营养标签（Transparency Cards）的框架设计" query="请详细拆解 Autodesk 等企业使用的 AI 营养标签（Transparency Cards）应该包含哪些核心字段与治理指标。"/>
<Elicitation label="探讨 Agentic Workflow 在企业合规与安全问卷中的应用" query="如何利用 Agentic Workflow（智能体工作流）自动化处理企业级安全与合规问卷？请给出架构思路。"/>
</ElicitationsGroup>
## Notable quotes
> [00:04](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=4s) 信任是增长的驱动力，而 AI 治理是 AI 创新的发射台 — Navrina Singh
> [00:42](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=42s) 信任绝非虚无缥缈的情感，而是完全可以进行科学量化的指标 — Kate Adams
> [16:40](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=1000s) 企业 AI 信任不仅仅是客户信任，它就是业务本身 — Kate Adams
> [16:40](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=1000s) 你是否清晰了解 Autodesk 正在如何使用你的数据以及为什么使用？ — Kate Adams
> [16:40](https://www.youtube.com/watch?v=OpOCHJLCMw8&t=1000s) 我骨子里是个安全人员，我不信任开源 — Kate Adams
## Mentioned
Navrina Singh, Kate Adams, Credo AI, Autodesk, Ai4, Taylor Swift, Eras Tour, Deloitte, The Four Factors of Trust
## Video description
Kate Adam - Director, Customer Trust & Employee Awareness, Autodesk
Navrina Singh - CEO/Founder, Credo AI
As AI moves from experimentation to enterprise-wide adoption, trust cannot be treated as an afterthought. This fireside conversation between Navrina Singh, Founder and CEO of Credo AI, and Kate Adams, Director of Customer Trust & Employee Awareness at Autodesk, will explore how organizations can connect AI innovation with clear governance, transparency, and employee awareness. The discussion will examine both the human and operational dimensions of building confidence in AI, including how leaders can create shared accountability across teams. Attendees will leave with practical considerations for scaling AI responsibly while maintaining trust with employees, customers, and other stakeholders.
[Source](https://www.youtube.com/watch?v=OpOCHJLCMw8)
Source quotes
[0:04] Navrina Singh: 信任是增长的驱动力，而 AI 治理是 AI 创新的发射台
[0:42] Kate Adams: 信任绝非虚无缥缈的情感，而是完全可以进行科学量化的指标
[16:40] Kate Adams: 企业 AI 信任不仅仅是客户信任，它就是业务本身
[16:40] Kate Adams: 你是否清晰了解 Autodesk 正在如何使用你的数据以及为什么使用？
[16:40] Kate Adams: 我骨子里是个安全人员，我不信任开源
Source quotes
[16:38] Kate Adam: How clearly do you understand what Autodesk is doing with your data and why?
[16:38] Kate Adam: I'm a security person at heart... I don't trust open source
Original transcript
[0:04] I'm Nina Singh. I'm the founder and CEO
of Credo AAI. We are the leading provider
[0:08] of AI governance and trust platform. And I am so
thrilled to have this conversation not just with
[0:15] uh one of our very special customers, Autodesk,
uh who really are proving that trust is a growth
[0:22] driver and AI governance actually is a launchpad
for AI innovation. but also because Kate here is
[0:29] a swifty and I am so thrilled because as a massive
swifty I just got some trust bracelets. So after
[0:36] the session if anyone is interested in trust
bracelets you know where to find Kate. So Kate
[0:42] welcome so excited to have you here. Thank you
Nina. Yes, I am a Swifty and I could not stop
[0:49] making friendship bracelets after the ARS tour
ended and so I started making Autodesk trust
[0:56] friendship bracelets and then it it's a problem
and so I need them to be productive. So yes,
[1:02] please come see me if you'd like a friendship
bracelet. Absolutely. And I think that just
[1:06] basically highlights uh Kate's very special role
in Autodesk. So as the director of customer trust,
[1:13] you know, I think Kate has one of the hardest
jobs in AI right now because to get frontier class
[1:19] capability honestly is very easy. The capability
hang is real. What is more critical and real right
[1:26] now in enterprise AI is the trust overhang that
we are seeing and that trust overhang is literally
[1:32] causing one enterprises to not adopt NFAI not even
deploy enough but it's also preventing them from
[1:40] becoming AI native. So Kate very curious what
does director of customer trust mean? That's a
[1:46] great question. Um I didn't know what it meant
before I took the role. Um but essentially the
[1:53] role was created because in an you know enterprise
software company especially with uh products in
[2:01] the cloud and developing 3D geometric AI models
uh for the design and make industry. We have a lot
[2:09] of customers who want to ensure that they that
Autodesk can protect their data that Autodesk
[2:18] can protect their confidentiality um of their
data and you know basically all of the things
[2:24] that come with trusting a company with terabytes
of data um and you know some very confidential
[2:30] highly sensitive uh information uh designs and
drawings and manufacturing processes and the next
[2:38] big movie. Um, and so my team, my role was created
to basically be that bridge between the internal
[2:46] trust organization which is comprised of infosc,
application security, privacy, responsible AI,
[2:54] resilience, governance, risk and compliance, um,
and what we call trust architecture. um basically
[3:01] all of these different disciplines that overlap a
lot um and are ne necessary for companies of all
[3:10] sizes from you know your Fortune 100 all the way
down to your mom and pop shops um to really you
[3:17] know get past that hurdle of you know what about
my data um and move on to the actual benefit that
[3:25] they are they are wanting to see out of a design
and make uh software platform So my team talks to
[3:31] customers every single day. Um we also own our
trust center and um you know we really really
[3:40] we're the face of trust. Um we also take a lot of
the customer trust use cases um because software
[3:47] can't be you have to have security and privacy
use cases um governance use cases within the
[3:54] product itself because we our customers have trust
organizations. they need to govern their software
[4:00] stack. Um, and so we take those use cases and
we feed that back to the product teams. We work
[4:05] with them to get them implemented. Um, and so
it's really this great feedback loop where we
[4:10] actually have the customer sentiment and their
anecdotes and their use cases and their, you know,
[4:14] the the the real data that drives business change
internally. Um, and uh, the way that that has
[4:20] worked out has been really rewarding and fun and
the challenges are ever present, especially now.
[4:27] Um, but I I I think I manifested this job. I
love it. Oh, that's fantastic. Uh, you know,
[4:33] I'm just very curious with a show of hands, how
many of you actually have used or heard the word
[4:40] AI trust in the past 24 hours at the conference.
Great. I'm surprised for the remaining because I
[4:47] thought every booth out there is uh really saying
that their solution can help you build trusted AI.
[4:53] So I think one of the things we really want to
get a little bit granular on Kate is measurable
[4:58] trust. Your customers are designing buildings,
aircrafts, you know, structures that's going
[5:03] to last 30 plus years. It's not just a workflow
or marketing application. So I'm very curious,
[5:09] what does measurable trust in Autodesk mean?
And more importantly, how are you making that
[5:14] real for your customers? This is a topic I could
talk about all day, but I see the timer there.
[5:19] Um but trust actually is very measurable. Um
we do quarterly surveys. We have we base our
[5:28] uh surveys on a methodology uh actually from
Deote. Uh two women from Deote wrote a book called
[5:35] the four factors of trust and they did a lot of
research on actually understanding how individuals
[5:41] define trust. Is there a common definition and
how can you measure that? because trust, loyalty,
[5:50] um you know, renewals, revenue, they're all tied
together. Um and so, you know, when we talk about
[5:58] measuring trust, we survey our customers every
quarter. We measure the four factors. Humanity,
[6:04] so caring. Do we care about them beyond their next
PO? Transparency. Do they understand what we're
[6:10] doing, how we're doing it? Do they feel like
they know what's going on at any given time?
[6:16] capability. Are our products capable? Do
they feel that we as a company are capable
[6:21] of delivering what we've committed to? And the
last pillar is reliability. Can they rely on us
[6:26] to do what they we say we're going to do when
we say we're going to do it consistently? Um,
[6:32] and so we measure it's it's a it's a quantifiable
metric. we measure those four pillars across all
[6:37] of our segments and that really tells us where we
need to maybe focus a little more or where we're,
[6:44] you know, where our hypotheses were correct that
this bet actually paid off. Um, and one of the the
[6:51] the piece that my team is particularly focused on
is transparency. What are you doing with my data?
[6:58] How are you using it? How are you securing
your cloud products, your software, right?
[7:03] What are your models doing? Right? How do I know
that my data is going to be protected from point
[7:10] A to point B? How do I know that it's it's not
going to be uh used in ways I don't like? Um,
[7:16] and so my team, customer trust, is really focused
on, you know, we we pay attention to these metrics
[7:22] quarter over quarter. Um, but we also, you know,
we're actioning those scores. we're actioning what
[7:30] needs to change to get those scores up and to
really become have Autodesk become synonymous
[7:36] with a trustworthy company. And I think one of the
things that we have seen in the past six years of
[7:42] creating this AI governance category where AI
governance is the how and measurable trust is
[7:47] the outcome is in the long run enterprises that
are investing in the infrastructure of trust are
[7:54] the companies that are going to win. It's not
going to be just how quickly can can you adopt
[7:59] a capable model, but the key question is is that
capable model actually going to drive outcomes
[8:05] for you whether those are internal outcomes or
whether those are customerf facing outcomes.
[8:09] So I'm very curious like the four tenants that you
just mentioned, humanity, capability, reliability,
[8:15] and transparency. Did I get that right? Yes. Yes.
Great. I'm paying attention which is great. Uh you
[8:20] know, one of the things that we've been very proud
to see from Autodesk is your transparency cards.
[8:25] So maybe just talk to us like why you know an
artifact that you can actually share with customer
[8:30] actually builds that trust and and how are you
making those transparency cards a reality. Yeah.
[8:36] So you know when we we started on this journey of
not only building AI but adopting AI internally um
[8:43] at an accelerated pace um you know we quickly ran
into an obstacle right where we didn't have that
[8:52] transparency we were you we could use ourselves as
an a use case our internal teams for our vendors
[9:00] said hey we want to know what is going on behind
the scenes we we don't like we use black boxes,
[9:06] we want to know what's going on with Autodesk
data. And so we took that model um and we applied
[9:12] that to our customers based on, you know, what we
wanted, empathy, but also what our customers were
[9:19] telling us they wanted. And so we knew we needed
to surface, you know, what models are being used,
[9:27] what the feature is really for, how is data being
safeguarded, what are the data sets being used,
[9:33] um, you know, is there encryption, is there are
there guard rails, where does the training happen,
[9:39] where does the inference happen? um we needed
to surface that on a feature by feature level,
[9:44] but we needed to do so in a way that wasn't a
block of text on a web page because nobody wants
[9:50] to read that. Um and we have 5 billion products
and all of them have various AI features. And
[9:57] so we decided we were going to use something
that I think everybody here is familiar with
[10:03] it and could identify it on site, which is
nutrition labels. You see a nutrition label,
[10:08] you know exactly what you're going to
get. you're going to get the facts,
[10:11] the nutrition facts that are going to inform
you on whether you want to eat this thing. Um,
[10:18] you have all the information. And so we did
that for our AI features. We put them in the
[10:23] form of a nutrition card uh for every single
feature and um posted them on our trust center.
[10:32] And that has helped immensely because one, we are
we're putting the data out there in a way that is
[10:40] easy to identify and in a way that is very clear
about what's going on. And no, they're not static,
[10:49] right? We're evolving them and and as our
customers in the market signal to us they want
[10:54] different kinds of information or some, you know,
deeper on one thing or not as deep on the other,
[10:59] um, we evolve them. Um, but it allows us to to
be as transparent as our customers want us to
[11:07] be in a way that isn't a heavy lift um and in
a way that um you know is is actually kind of
[11:14] a very cool visual as well because they're
super recognizable um and you know our own
[11:22] internal teams have have come to kind of expect
that oh I make a new AI feature going to need a
[11:27] transparency card great so you know The session's
title is building trust in enterprise AI. And I'm
[11:33] a big believer you can't build you know trust
in enterprise AI if you yourself are not using
[11:39] AI and you know the panel before this or the
discussion before this was about how AI is
[11:44] getting you fired which certainly is a sentiment
among lot of companies but what has been really
[11:49] exciting to watch Autoes transformation from the
outside is how they are becoming AI native. So I'm
[11:55] very curious Kate like talk to us specifically
what does AI native in Autodesk mean and more
[12:02] specifically give me instances of when you were
trying to be transparent in that process and maybe
[12:07] that rigor didn't you know meet the engineering or
the AI builder demands very curious to hear your
[12:13] thoughts and this is a very timely question
because I was literally in a leadership off
[12:20] site two weeks ago discussing this very thing
and these same questions came up from my peers
[12:24] Um and and what we we kind of landed on after a
week of huddling and you know trying to figure
[12:30] out where we all were with our own use of
AI internally at Autodesk is that AI native
[12:36] doesn't mean you're using AI for everything. AI
native means that you are recognizing that you
[12:43] have another powerful tool in your tool set right
and to um use it when it's appropriate. It's not
[12:51] appropriate for everything. like you don't use
a hammer for every single thing you need to do
[12:54] in construction. It is has a specific use case
and it's very good at what it does. Um but it
[13:00] is learning an AI native is learning experimenting
with AI on different use cases testing hypotheses.
[13:09] Um and I think the most important piece uh is
moving from a uh an operational model where you
[13:20] are going from basically an agile methodology
agile methodology to an AI native method
[13:27] methodology where you are shipping with 20 to 30%
improvement. you're not going for perfect because
[13:38] we don't have time for that anymore. We have to
innovate so quickly. We need to fail fast and that
[13:44] means being okay with experimenting, taking time
to experiment because slowing down will allow us
[13:50] to speed up um and really changing our mindset
as leaders, changing our mindset as a company,
[13:58] our operational model to fully utilize the tool
set. Move fast. breaking things is good as long as
[14:04] you learn something and being extra communicative
about what has worked and what hasn't worked. So
[14:10] specifically um my team also does we we work on a
lot of the customer security questionnaires in the
[14:18] procurement process and so we we had this issue
where you know we're using this tool and it was
[14:24] a very manual process as you can imagine we have
again I said 5 billion that's hyperbole but it's
[14:30] a lot of products we have a lot of products at
Autodesk and there are a lot of nuances and so it
[14:35] took a lot of time to get to the answers we needed
because we would go to a person and that person
[14:41] would take a couple weeks to figure it out did
not scale. And so we had this hypothesis, my team,
[14:48] where we said, well, you know, we we we're getting
very used to this agentic workflow. Let's see if
[14:54] we could ask the product itself, right? Let's
can we ask the code what it does? Does it do
[15:01] this thing? And um we tested that, ran some tests
and you know a lot of human intervention but we we
[15:10] operated on this 20 to 30% incremental improvement
and we're now 4 months into this experiment and we
[15:18] have been able to complete these questionnaires
and with higher accur accuracy and we've been
[15:25] able to reduce the time it takes by 5x which is
crazy. and we're able to do it without bothering
[15:33] the engineers at all because we're literally
just asking the product. Um, and doing that has
[15:39] also led to other use cases where yeah, ask the
product yields white papers and uh understanding
[15:49] components uh better and more accurately and
understanding dependencies more accurately and
[15:55] more quickly and mapping them out. And so that's
a success case. But the other piece is what does
[16:00] failure look like? Right? If you're going for 20
to 30%, you're going to fail very quickly if that
[16:06] failure is in your future and you can pivot and
go an alternate route. And that might take you
[16:12] a couple days. But that's better than trying to
go for perfect and then realizing that it's not
[16:18] working. And two months have passed. Two months is
a lifetime, especially in this world where I mean,
[16:25] we were just talking. No, a year ago, no one
MCP didn't exist, right? So, it's it's really
[16:32] a mindset shift and really using those tools
effectively for the use cases in which they can be
[16:38] used. So, we didn't even get to the details of AI
governance or what does measurable trust metrics
[16:45] look like? Um, how are you operationalizing that
at Autodesk? We're already at 3 minutes only left.
[16:52] So, I think we're going to do a quick rapid fire.
Kate um and and keep these short. These are just
[16:57] like the first thing that comes to your mind. So
one metric you would put on a slide that everyone
[17:05] from privacy AI builders AI data science and trust
organization align on that is a good measure of
[17:12] you know transparency. We measure this. We have a
question we ask all of our customers. How clearly
[17:20] do you understand what Autodesk is doing with your
data and why? I put that on every slide. Okay. Is
[17:26] yeah that that's the metric I would use. Great. Uh
what is one thing you will tell uh you know folks
[17:33] here who might be running enterprise AI governance
functions that they should not be doing? That is
[17:41] a really good question. What should you not be
doing? Um, you know, I think because we're all
[17:48] in this journey to AI maturity, I think that we
should not be assuming that we're all that we're
[17:57] all behind in this maturity journey. Yes, there
are people who who are eons ahead, but the vast
[18:05] majority of companies um especially, you know,
Fortune 500 outside, we're all in this together.
[18:12] And so we should not be we should not be thinking
that a we know everything or b that we're so far
[18:21] behind that there's no hope. We should be learning
from each other. Even in competitive situations,
[18:28] we're going to be benefiting from sharing stories
of failure and success and and taking the things
[18:33] that work and applying them to our own businesses
and learning from the things that didn't work for
[18:38] other people and saying, "Okay, maybe I don't want
to do it that way." Um, so let's not work in silos
[18:44] even across competitive situations and industries.
What do you trust more foundation model providers
[18:52] or open source? I'm a security person at heart.
So this is a tough question. I don't trust open
[18:58] source. Wow. Okay. Uh, we have some conversations
to be had. And then last question. Uh,
[19:05] enterprise AI trust is not complete the sentence.
Enterprise AI trust is not is not only customer
[19:15] trust. Enterprise AI is business. Enterprise AI is
how you're using AI tools internally. How you're
[19:26] building and developing AI. Uh how you're putting
AI into your product set and the use cases that
[19:34] you're building AI for. all of it together in a
way that you can see and monitor and govern and
[19:41] pivot quickly. Awesome. Well, thank you so much,
Kate. You know, there's so much to uncover in
[19:48] what trusted uh enterprise AI means. We didn't
have enough time, but if you have questions,
[19:52] please stop by and uh give a big round of applause
to Kate. Thank you, Kate. Thank you so much, Nina.
