No Priors: AI, Machine Learning, Tech, & Startups

当AI智能体能够自主设计实验、收集数据并持续优化，而无需人类介入时，会发生什么？安德烈·卡帕西与莎拉·郭展开对话，探讨模型现状、工程与教育的未来、对就业市场的影响，以及他的项目AutoResearch——在该项目中，智能体能够自主完成AI研究的闭环（包括实验设计、模型训练与优化）。 00:00 安德烈·卡帕西介绍 02:55 当前能力边界何在？ 06:15 编程智能体的精通形态 11:16 自然语言编程的二级效应 15:51 为何选择AutoResearch 22:45 AI时代的相关技能 28:25 模型特化 32:30 构建更多人类与AI的协作界面 37:28 就业市场数据分析 48:25 开源与闭源模型之争 53:51 自主机器人技术 1:00:59 MicroGPT与智能体教育 1:05:40 结语

吴恩达始终处于快速演变的AI技术前沿，创立了Google Brain、AI Fund和DeepLearning.AI等公司和项目。因此他比任何人都更清楚，那些在2025年仍沿用2022年运营方式的创始人正在犯错。Sarah Guo和Elad Gil与这位AI革命教父展开对话，探讨智能体AI的崛起，以及这项技术如何彻底改变了从成功创始人的特质到小团队价值等一切要素。他们讨论了未来能力增长的潜在来源、模型自我进化的可能性，以及吴恩达为何不喜欢"氛围编程"这个术语。此外，吴恩达还论证了组织中的每个人——而不仅仅是工程师——都应该学习编程的理由。 每周订阅新播客。发送邮件至show@no-priors.com反馈意见 关注我们的Twitter账号：@NoPriorsPod | @Saranormous | @EladGil | @AndrewYNg 章节： 00:00 – 吴恩达介绍 00:32 – 能力增长的下一个前沿 01:29 – 吴恩达对智能体AI的定义 02:44 – 构建真正智能体的障碍 06:09 – 智能体AI的前沿领域 08:12 – 模型会自我进化吗？ 09:05 – 氛围编程 vs AI辅助编程 09:56 – 氛围编程是否在改变创业本质？ 11:35 – 加速项目管理 12:55 – 成功创始人画像的演变 19:23 – 寻找优秀产品人才 21:14 – 为单一用户画像还是多用户画像构建 22:47 – AI时代领导者与团队的必备条件 28:21 – 保持小团队的价值 32:13 – 下一个行业变革 34:04 – 投资公司与孵化器的自动化未来 37:39 – 技术型人才首次创业 41:08 – AI未来五年的广泛影响 41:49 – 结语

What happens when you put code-generation agents in the hands of enterprise IT teams? With Serval, you get a system where it's easy for everyone to be a builder. Sarah Guo and Elad Gil sit down with Serval co-founder and CEO Jake Stauch to analyze his company’s direct challenge to ServiceNow as the enterprise automation and service management system of record. Jake explains that by building an AI platform for enterprise IT departments with Serval, users can safely build while still safeguarding permissions by strict deterministic approvals and an audit trail. He also talks about ProServal, the company’s team that embeds automation engineers directly with customers in order to provide ongoing support and service. Jake also discusses how AI was the big unlock to taking on ServiceNow, why he’s betting on coding agents getting even better, why consumer is a window into what will happen next in enterprise and why the SaaS-pocalypse might look more like a consolidation, why background agents are still a moonshot in enterprise software, and the importance of hiring product-oriented engineers. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @jakeserval | @getserval Chapters: 00:47 – Jake Stauch Introduction and Serval’s Background 03:00 – Expanding Coding Agents Beyond Engineers 04:40 – Limitations with Personal Agents in Enterprise 06:29 – Deploying Serval Effectively in Large Orgs 07:55 – Day 1 Ambition and Attacking Enterprise Systems of Record 09:51 – Taking on ServiceNow with Workflow Automation Platform 11:01 – Betting on the Evolution of Coding Agents and Building Guardrails in Enterprise 14:30 – Big Risks, Big Rewards, and Big Predictions for Serval and Coding Agents 17:24 – Adopting Different Sets of Building Principles at an Enterprise Level 21:31 – Token Maxing, Long-Term Budget Impacts, and ROI 23:21 – The SaaS-pocalypse and SaaS vs. AI 25:21 – The Mental Model for Delivering Platform-Level Power to the IT User at Enterprise 27:26 – Building Automations with Catalyst and Raising Technical Competence 29:38 – The Next Wave: Background Agents and Autonomy in Enterprise 33:07 – Hiring Velocity and Unlimited Ambition at Serval 34:59 – Keeping Pace with Customers and Fundamental Shifts in Product Function 37:46 – Conclusion

The pyramid of pain, from the least to the most dangerous attackers: Base: hacktivists (most numerous) Middle: cybercriminals Top: nation state actors (fewest) Now AI is closing the gap. CrowdStrike CEO George Kurtz discusses a new type of threat: the agent state

27 seconds. That’s the fastest CrowdStrike has seen an attacker move from one system to another. It used to take hours or days. Here’s what George Kurtz says changed.

In the transition to the agentic web, security has become an afterthought. Now, organizations are racing to figure out what AI security means, and how to keep their data safe. Live from Cloudflare Connect, Sarah Guo sits down with Cloudflare CTO Dane Knecht to discuss the changing attacker landscape, and how his company is allocating its engineering resources to meet this emerging security challenge. Dane explains how enterprises that have moved away from castle-and-moat infrastructure to more of a zero trust infrastructure for security are in better positions than those who haven’t. He also discusses the challenges of patching 70 years of software and why the software development lifecycle will need to change for security for the agentic web, why he sees token spend in the next 18 months primarily going towards security rather than engineering, how the current web is in a ‘messy middle’ state where humans and agents are riding the same ‘rails’ on the internet, Cloudflare’s conference monetization gateway release as well as its Signals project with OpenAI, the rationale behind the release of Cloudflare OS, and the thinking behind a recent company restructuring. Plus, why Cloudflare hired one thousand interns, and why curiosity about AI is required to be successful at Cloudflare. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @Cloudflare | @dok2001 Chapters: 00:00 – Trailer 00:55 – Cloudflare CTO Top Priorities 05:03 – Today’s Attacker Landscape 06:16 – State of Internet Infrastructure 07:31 – Shift Towards Security Spend 08:39 – Road to Better Software and Internet Security 09:50 – Building Secure Production Software 11:41 – Reducing Surface Area 12:54 – Near Future of the Web 17:27 – Monetization and Signals 19:13 – Success for Content Creators 20:07 – Workplace OS and Use Cases 24:07 – Restructuring for AI-First 27:11 – Token Budgeting 29:32 – Cloudflare’s Intern Strategy 33:39 – Importance of Curiosity 35:29 – Attacking the Legacy Software Security Problem 37:08 – Engineering Team Prioritization 38:25 – Open vs. Closed Models 39:43 – Democratizing Security for Everyone 41:29 – Conclusion

Adversaries are already running AI agents, and the fastest attacks now play out in seconds. Sarah Guo and Elad Gil sit down with CrowdStrike president and CEO George Kurtz on what AI-equipped attackers mean for enterprise security. George explains how agents are flattening the adversary pyramid by putting advanced tradecraft in the hands of low-skill attackers, and why CrowdStrike and Nvidia are building a frontier-style lab of their own to train security models. They also get into how open-weight and closed models compare on security, where CrowdStrike is directing its platform investment, lessons from the recent Hugging Face incident, and why George sees AI, like every platform shift before it, as a net creator of opportunity. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @CrowdStrike | @George_Kurtz Chapters: 00:00 – Trailer 00:59 – George Kurtz Introduction 01:24 – Agentic Landscape 02:45 – AI and Adversaries 04:51 – Open Weight Models and Adversarial Attacks 06:37 – Adversary Geography 08:34 – Attacker Capability in Context 11:24 – Hugging Face Takeaways 13:59 – CrowdStrike Frontier Security Models 16:32 – CrowdStrike Key Offerings 19:51 – What Differentiates CrowdStrike’s Models 23:57 – Near Term State of Security World 25:46 – State of Model Distillation 27:53 – Open Weight Model Importance and Cost 29:59 – Assessing Real Agent Threat 32:40 – Areas to Watch in AI 35:02 – Case for Defense Dominance 39:15 – Executive Advice 43:16 – Shifts Create Opportunity 45:27 – Conclusion

NVIDIA packs six to nine custom chips into one powerful system. Fractile CEO Walter Goodwin on how chip companies can catch up: keep a rolling frontier of bets ready to ramp.

OpenAI reassigned 25% of its production engineers to security. Then it used its top AI model to find vulnerabilities in its own systems. President Greg Brockman explains why.

First party chips are mostly about getting a better price from NVIDIA. Fractile’s Walter Goodwin on why there might be some truth to that joke:

OpenAI president Greg Brockman’s thoughts on our No Priors conversation: “Security spend should actually vastly outstrip software spend.”

Technology has always been the fundamental driver of improvement for human quality of life. And with the recent rollout of advanced frontier models, like OpenAI’s GPT-6, humanity can now massively accelerate scientific discovery for its benefit. Sarah Guo and Elad Gil are joined by OpenAI President and co-founder Greg Brockman to discuss the company’s rapid progress across frontier models, cybersecurity, and scientific discovery. Greg shares how nearly 400 century-old math problems have been solved using its AI agents, and how the technology has the potential to transform other fields like biology in the realm of drug discovery. He also talks about how proactive, always-on AI agents like Dots impact modern productivity, why he believes the best UI is no UI at all, how cybersecurity is in a critical window of time to use peak intelligence to find and fix vulnerabilities in systems, and why OpenAI remains committed to keeping humans in the driver’s seat even as models surpass us in capability. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @gdb | @OpenAI Chapters: 00:42 – Trailer 00:59 – Greg Brockman Introduction 01:11 – AI and Advancements in Mathematics 05:55 – AI Communication and Oversight 07:45 – AI for Overcoming Biology Data Bottlenecks 11:28 – Revolutionizing Drug Discovery 14:18 – Changing Regulations to Accelerate Progress 16:51 – Dots the Proactive Agent 19:19 – Unifying Work and Consumer Agents 22:02 – How UI Will Evolve 26:08 – Cybersecurity in the AI Age 32:55 – Shift in Security Investment 35:49 – OpenAI’s Human-Centric Ethos 38:28 – Conclusion

"We can now have extreme intelligence that is guiding our experiments, but now we're just experiment bottlenecked.” Fractile CEO Walter Goodwin on why smarter AI still can’t make chip design instant.

With 90% retention, startups almost can't compete in insurance brokerages. Sequence Holdings bought into one instead. CEO Michael Lee on the Baldwin thesis. New No Priors episode, out now.

Why doesn't every frontier lab build its own chip? Because if a rival finds a 5x efficiency gain that only runs on their silicon, you're out of the race before your chips deploy. Fractile CEO Walter Goodwin on why third party chips survive.

You can't separate cyber offense from defense. Reflection CEO Misha Laskin explains how restricting access to powerful AI capabilities can leave defenders less equipped to protect themselves, while open models can help level the playing field.

Is the future of frontier AI open or closed? ReflectionAI co-founder and CEO Misha Laskin joins Sarah Guo and Elad Gil to discuss the launch of Beam, the company’s 500 billion parameter open-weight reasoning model. Misha breaks down the pre-training and reinforcement learning required to produce Beam’s reasoning efficiency, the shift in enterprise compute from renting to owning intelligence, and how he believes that open models will capture the majority of global token demand. He also talks about the open model ecosystem in China and why competition with China’s models is a good thing, safety considerations around open models, and how frontier-level open models may accelerate the pace of scientific discovery. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @MishaLaskin | @reflection_ai Chapters: 00:42 – Misha Laskin Introduction 00:59 – Latest with ReflectionAI 02:07 – Challenges Building an Open Model 03:56 – ReflectionAI’s Agentic Shift 07:37 – Resources for Model Training 10:09 – Scaling Efficiency 12:18 – Training Beam 16:51 – Where Model Value Comes From 19:55 – Beam’s Reasoning Efficiency 22:35 – Monetizing Open Weight Models 25:03 – Future of Open Versus Closed Tokens 31:29 – Competition with Open Models 34:38 – Chinese Open Source Model Ecosystem 38:40 – Will Chinese Models Remain Open 44:44 – Safety and Open Models 53:02 – Debating Access to Powerful Tools 57:16 – AI and Scientific Progress 01:00:36 – Data Centers and Jobs 01:02:07 – Beam and Scientific Research 01:05:37 – Research Head Count to Compute Ratio 01:10:29 – Conclusion

Is the future of frontier AI open or closed? ReflectionAI co-founder and CEO Misha Laskin joins Sarah Guo and Elad Gil to discuss the launch of Beam, the company’s 500 billion parameter open-weight reasoning model. Misha breaks down the pre-training and reinforcement learning required to produce Beam’s reasoning efficiency, the shift in enterprise compute from renting to owning intelligence, and how he believes that open models will capture the majority of global token demand. He also talks about the open model ecosystem in China and why competition with China’s models is a good thing, safety considerations around open models, and how frontier-level open models may accelerate the pace of scientific discovery. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @MishaLaskin | @reflection_ai Chapters: 00:05 – Misha Laskin Introduction 00:22 – Latest with ReflectionAI 01:30 – Challenges Building an Open Model 03:19 – ReflectionAI’s Agentic Shift 07:00 – Resources for Model Training 09:32 – Scaling Efficiency 11:43 – Training Beam 16:14 – Where Model Value Comes From 19:18 – Beam’s Reasoning Efficiency 21:58 – Monetizing Open Weight Models 24:28 – Future of Open Versus Closed Tokens 30:54 – Competition with Open Models 34:01 – Chinese Open Source Model Ecosystem 38:03 – Will Chinese Models Remain Open 44:07 – Safety and Open Models 52:25 – Debating Access to Powerful Tools 56:39 – AI and Scientific Progress 59:59 – Data Centers and Jobs 01:01:32 – Beam and Scientific Research 01:05:03 – Research Head Count to Compute Ratio 01:10:18 – Conclusion

AI's real opportunity isn't speed. It's rethinking how a company works. Sequence Holdings CEO Michael Lee on why this is an industrial revolution moment.

What does it mean to "train" an AI model? Inception Labs CEO Stefano Ermon explains, it's compression. The model finds the most efficient way to compress the data, and that's how it learns structure.

"We're trying to do one deal per year. That's it." Sequence Holdings CEO Michael Lee on why his firm skips the typical PE playbook entirely.

AI is getting faster at solving hard problems. But what happens when the experiments needed to test those solutions become the bottleneck? Fractile CEO Walter Goodwin explains why accelerating experimentation is becoming just as important as accelerating AI itself, from chip design to scientific discovery.

Founder and CEO of full-stack AI chip company Fractile, Walter Goodwin, joins Sarah Guo to discuss the bets he’s made on the future of the chip market as other major players like Broadcom, NVIDIA, and AMD try to accelerate their workloads. They discuss the difference in Fractile’s newer approach on model architecture with a full-stack team in the current chip landscape and the technical bets they’re making in that direction. Walter also talks about compressing the gap between the chip design cycle and its payoff period, and making a generational leap in AI inference to realize the bet in volume against the value to be captured. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @goodwin_ml | @fractile_ai Chapters: 00:46 – Walter Goodwin and Fractile Introduction 02:29 – The Chip Landscape Now 04:56 – Common Handoffs From Architecture-Focused Players 07:18 – Full Stack Approach and Team Setup 09:49 – Fractile’s Most Important Technical Bets 15:32 – Workload Predictions and Compressing the Chip Design Cycle 23:03 – Architect Intent to Output Bottlenecks and Accelerating Trials 28:16 – Workload Bets on Model Architectural Shifts 31:20 – The Future of AI Chip Players and Market Structure 35:14 – Conclusion

"We were able to generate text 10x faster compared to the autoregressive model." Inception Labs CEO Stefano Ermon on the 2024 breakthrough paper that proved diffusion could match traditional LLMs on quality, and why that was enough to start a company.

Services firms optimize to keep you as a customer, not to reinvent your business. Sequence Holdings CEO Michael Lee on the incentive problem behind the entire services industry.

Blackstone celebrates the investor. Palantir celebrates the engineer. Sequence Holdings CEO Michael Lee on No Priors: who a company celebrates decides who it can attract.

Can AI transform legacy incumbents rather than replacing them? Sequence Holdings co-founder and CEO Michael Lee joins Sarah Guo to discuss how holding company structures and engineering integrations are reshaping market leaders from the inside out. Michael details Sequence’s $7.7 billion take-private transaction of Baldwin alongside Dell Family Office (DFO), and shares his thesis on why traditional consulting models and software sales fall short for real enterprise AI transformations. They also talk about why permanent holding company structures are good for long-term compounding, real-world results from applying frontier engineering to BankSouth, and Michael’s lessons from his time in public investing, private equity, and operating at the intersection of market incumbents and AI. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @mjlee_2014 | @seqholdings Chapters: 00:34 – Michael Lee Introduction 01:03 – Sequence Holdings and Baldwin 01:54 – Idea for Sequence 04:36 – Incumbents in the AI Era 06:32 – Why a Holding Company 11:21 – Recruiting Top Engineers 13:08 – Investing in BankSouth 17:46 – Why an Insurance Brokerage 20:17 – Atlas Platform Explained 23:53 – Traditional Private Equity Limitations 27:23 – What Sequence Looks For in Management Teams 31:04 – Accomplishments at BankSouth 34:45 – Founder Lessons 36:08 – Story of Dell Partnership 37:10 – Career and Investment Approach 40:00 – Value of Exceptional People 42:23 – Conclusion

"At some point it'll be abnormal to have children without genetic screening and editing." "It's like driving without a seatbelt." AI is forcing us to take control of human evolution. Brian Armstrong explains:

Today's LLMs can't generate token 10 until token 9 exists. Inception Labs CEO Stefano Ermon thinks he's found the winner: diffusion LLMs that generate tokens in parallel. New episode of No Priors, out now.

As generative AI hits hardware and latency bottlenecks, Stanford professor, diffusion pioneer, and Inception co-founder and CEO Stefano Ermon is betting on a radical new architecture. Stefano joins Sarah Guo to talk about Inception, and how his team is applying diffusion architecture beyond images and video into discrete text and code generation. Stefano explains the limitations of autoregressive LLMs, as well as why parallel token generation in diffusion models offers superior inference scaling and hardware utilization on standard GPUs. He also shares details about Inception’s Mercury models, real-world voice agent applications, the software stack required to serve diffusion-based models at scale, academia’s role at the frontier of AI innovations, and why the next era of AI competition will be defined by efficiency. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @StefanoErmon | @_inception_ai Chapters: 00:00 – Stefano Ermon Introduction 00:35 – Research Background 02:54 – Starting Inception 05:59 – Why Diffusion Beats Autoregressive 11:10 – Discrete vs. Continuous Modalities 13:19 – Inception Today 16:45 – Where Speed Wins 17:31 – Inception Customer Base 18:49 – Interaction with Hardware Landscape 19:34 – Inception and the Broader Industry 21:41 – Data Compression and Structure 24:45 – Controllability of Diffusion Modeles 27:25 – Emergent Capabilities at Scale 29:02 – Future Workload Split Between Diffusion vs. Traditional 30:03 – Adoption Challenges 31:44 – Hiring and Team Organization 32:50 – Recursive Self Improvement 34:02 – Resource Allocation 35:10 – Impact of Academia 38:13 – Conclusion