AI’s third era: the rise of persistent AI coworkers | Tara Seshan (OpenAI’s product lead)
Tara Seshan, OpenAI's product lead for ChatGPT and Codex, discusses how AI is fundamentally changing product management and knowledge work. She emphasizes the shift from theoretical planning to rapid empirical iteration, the importance of maintaining human ambition and opinionation in an age of AI capabilities, and how products like work mode and sites are enabling persistent AI coworkers.
Summary
Tara Seshan shares insights from her year at OpenAI leading product for both ChatGPT and Codex. She begins by describing three eras of AI products: chat-based interfaces, agent-based systems, and emerging persistent AI coworkers. A key theme throughout is the challenge of building for a rapidly moving target—she emphasizes that building for current model capabilities, past capabilities, or predicted future capabilities are all equally wrong, and the optimal approach is building on a two to three month horizon.
Seshan discusses how her role as a PM has fundamentally shifted from the theoretical, document-heavy approach she used at Stripe to a more empirical, prototype-driven methodology. Rather than writing lengthy reasoning documents, she now focuses on quickly testing hypotheses with users. She stresses that the core PM job—identifying the eigen question, testing it, and refining hypotheses—remains constant, but the speed and format have changed dramatically.
A surprising discovery at OpenAI was its openness compared to other companies she's worked at. Rather than a treasure trove of secret strategy, OpenAI operates as a founders-led organization where individuals act as founders of their product areas with minimal top-down direction. This creates a thin distance between product teams and the market.
Seshan emphasizes that in an era where everyone has access to the same AI tools, the competitive advantage shifts to human factors: ambition, opinionation, taste, and the ability to articulate a vision. She argues that the most effective people don't simply automate rote tasks but expand their range of possible capabilities—similar to how polymaths functioned before AI, but now accessible to everyone. A critical PM responsibility is elevating others' ambitions and reminding teams what's actually possible.
Regarding ChatGPT's product strategy, Seshan explains the current toggles between different modes (chat, work, Codex) as transitional states. The north star is removing these decisions entirely—users should simply describe their task and the system should automatically select the right interface and capabilities. Work mode represents the application of Codex's agent capabilities to knowledge work, not just coding.
Seshan draws an important distinction between coding and knowledge work: coding is output-oriented and verifiable through tests, while knowledge work requires understanding the process, reasoning, inputs, and citations. This means the product must evolve to show more collaborative elements, chain-of-thought transparency, and citations to help users validate results.
On writing, Seshan distinguishes between writing as thinking and writing as reporting. She automates the latter extensively but never automates the former, as the act of writing is how she clarifies her own thinking. At OpenAI, the shareable artifact has shifted from long documents to mocks, prototypes, and experimental results—a significant departure from her Stripe days.
Seshan discusses internal cultural memes at OpenAI: "Are we mainlining it yet?" (using the product all day), "Is this maximally accelerated?" (moving as fast as possible), and consciousness of AGI's implications. She shares specific product features like Sites (shareable, database-backed applications built through natural language) and Visualize (creating dynamic visualizations from data) as examples of how users can now accomplish previously impossible tasks.
On human value in an AI-augmented world, Seshan argues humans remain essential for accountability (owning outcomes), expression (the authorship and opinionation in software), and relational connection (caring for each other and elevating collective ambitions). She compares software to filmmaking rather than real estate—budget alone doesn't guarantee success; artistry and vision matter.
Reflecting on her time as a Teal Fellow and at Sutter Hill Ventures, Seshan notes that product-market fit, while important, is often preceded by product-marketing fit—the narrative and positioning can precede the product itself. She emphasizes that great PMs must shop their ideas around, invite critique, and strengthen them through collaborative iteration rather than presenting polished finished work.
About this episode
<p><strong>Tara Seshan</strong> leads product for Codex and ChatGPT Work at OpenAI (alongside previous podcast guest Andrew Ambrosino, who’s her engineering manager). Before OpenAI, Tara spent over six years at Stripe, where she joined as one of the first five product managers. She went on to lead product for Watershed, which <em>Time</em> magazine named one of the best inventions of 2022, and she is also a founder and Thiel Fellow. Most personally meaningful to me: Tara is one of the three inaugural Lenny’s Newsletter Fellows, a program I ran a couple of years ago to spotlight the most exciting up-and-coming product leaders.</p><p></p><p><strong>In our in-depth conversation, we discuss:</strong></p><p>1. The shift from “rowing” to “steering,” and why human judgment and ambition will become differentiators as AI takes on execution</p><p>2. How OpenAI thinks about building for model capabilities two to three months out</p><p>3. OpenAI’s best internal memes, such as “Is this maximally accelerated?” and “Are you mainlining it yet?”</p><p>4. Why ambition is the new bottleneck for companies, and why elevating others’ ambitions is now the key part of the PM job</p><p>5. Writing as thinking vs. writing as reporting</p><p>—</p><p><strong>Brought to you by:</strong></p><p><a href="https://workos.com/lenny" target="_blank"><strong>WorkOS</strong></a>—Make your app enterprise-ready, with SSO, SCIM, RBAC, and more</p><p><a href="https://mercury.com/command?utm_source=lennys&utm_medium=sponsored_newsletter&utm_campaign=26q3_brand_campaign" target="_blank"><strong>Mercury</strong></a>—Radically different banking, now with Command</p><p>—</p><p><strong>Where to find Tara Seshan:</strong></p><p>• X: <a href="https://x.com/tarstarr" target="_blank">https://x.com/tarstarr</a></p><p>• LinkedIn: <a href="https://www.linkedin.com/in/tarstarr" target="_blank">https://www.linkedin.com/in/tarstarr</a></p><p>• Newsletter: <a href="https://substack.com/@taraseshan" target="_blank">https://substack.com/@taraseshan</a></p><p>—</p><p><strong>Where to find Lenny:</strong></p><p>• Newsletter: <a href="https://www.lennysnewsletter.com" target="_blank">https://www.lennysnewsletter.com</a></p><p>• X: <a href="https://twitter.com/lennysan" target="_blank">https://twitter.com/lennysan</a></p><p>• LinkedIn: <a href="https://www.linkedin.com/in/lennyrachitsky/" target="_blank">https://www.linkedin.com/in/lennyrachitsky/</a></p><p>—</p><p><strong>In this episode, we cover:</strong></p><p>(00:00) Introduction</p><p>(02:18) What makes OpenAI’s culture so different</p><p>(06:42) Why AI product strategy is all about fast experimentation</p><p>(09:02) How the PM role is changing</p><p>(10:50) The shift from rowing to steering</p><p>(15:35) What changes when agents become coworkers</p><p>(20:05) Why ambition matters more than ever</p><p>(26:39) Building products for models that do not exist yet</p><p>(29:21) How ChatGPT’s Chat and Work modes differ</p><p>(34:01) How OpenAI ships so quickly at scale</p><p>(39:14) The vibe shift happening inside Codex</p><p>(42:20) Why traditional roles are beginning to blur</p><p>(45:59) Where humans will continue to provide unique value</p><p>(48:20) How Tara uses AI in her own work</p><p>(51:38) The magic of the /visualize command</p><p>(52:39) Writing to think versus writing to report</p><p>(57:10) How to use AI without losing your ability to think</p><p>(01:00:15) Tara’s biggest lesson from Sutter Hill</p><p>(01:04:16) ChatGPT’s site output</p><p>(01:05:01) Why knowledge work is becoming more like coding</p><p>(01:07:55) Lightning round and final thoughts</p><p>—</p><p><strong>Referenced:</strong></p><p>• Codex: <a href="https://chatgpt.com/codex" target="_blank">https://chatgpt.com/codex</a></p><p>• ChatGPT Work: <a href="https://openai.com/chatgpt-work" target="_blank">https://openai.com/chatgpt-work</a></p><p>• Stripe: <a href="https://stripe.com" target="_blank">https://stripe.com</a></p><p>• Watershed: <a href="https://watershed.com" target="_blank">https://watershed.com</a></p><p>• Thiel Fellowship: <a href="https://thielfellowship.org" target="_blank">https://thielfellowship.org</a></p><p>• Meet your Lenny’s Newsletter Fellows: <a href="https://www.lennysnewsletter.com/p/meet-your-lennys-newsletter-fellows" target="_blank">https://www.lennysnewsletter.com/p/meet-your-lennys-newsletter-fellows</a></p><p>• The rituals of great teams | Shishir Mehrotra of Coda, YouTube, Microsoft: <a href="https://www.lennysnewsletter.com/p/the-rituals-of-great-teams-shishir" target="_blank">https://www.lennysnewsletter.com/p/the-rituals-of-great-teams-shishir</a></p><p>• The nature of product | Marty Cagan, Silicon Valley Product Group: <a href="https://www.lennysnewsletter.com/p/the-nature-of-product-marty-cagan" target="_blank">https://www.lennysnewsletter.com/p/the-nature-of-product-marty-cagan</a></p><p>• Product management theater | Marty Cagan (Silicon Valley Product Group): <a href="https://www.lennysnewsletter.com/p/product-management-theater-marty" target="_blank">https://www.lennysnewsletter.com/p/product-management-theater-marty</a></p><p>• Patrick Collison’s examples of fast projects: <a href="https://patrickcollison.com/fast" target="_blank">https://patrickcollison.com/fast</a></p><p>• Inside ChatGPT: The fastest-growing product in history | Nick Turley (Head of ChatGPT at OpenAI): <a href="https://www.lennysnewsletter.com/p/inside-chatgpt-nick-turley" target="_blank">https://www.lennysnewsletter.com/p/inside-chatgpt-nick-turley</a></p><p>• Andrew Ambrosino on X: <a href="https://x.com/ajambrosino" target="_blank">https://x.com/ajambrosino</a></p><p>• Tyler Cowen’s website: <a href="https://tylercowen.com" target="_blank">https://tylercowen.com</a></p><p>• OpenAI’s CPO on how AI changes must-have skills, moats, coding, startup playbooks, more | Kevin Weil (CPO at OpenAI, ex-Instagram, Twitter): <a href="https://www.lennysnewsletter.com/p/kevin-weil-open-ai" target="_blank">https://www.lennysnewsletter.com/p/kevin-weil-open-ai</a></p><p>• “Chop wood, carry water” quote: <a href="https://buddhism.stackexchange.com/questions/15921/what-is-the-meaning-of-the-zen-quote-before-enlightenment-chop-wood-carry-wat" target="_blank">https://buddhism.stackexchange.com/questions/15921/what-is-the-meaning-of-the-zen-quote-before-enlightenment-chop-wood-carry-wat</a></p><p>• 4 questions Shreyas Doshi wishes he’d asked himself sooner | Former PM leader at Stripe, Twitter, Google: <a href="https://www.lennysnewsletter.com/p/shreyas-doshi-live" target="_blank">https://www.lennysnewsletter.com/p/shreyas-doshi-live</a></p><p>• Alan Kay: <a href="https://en.wikipedia.org/wiki/Alan_Kay" target="_blank">https://en.wikipedia.org/wiki/Alan_Kay</a></p><p>• Brie Wolfson on X: <a href="https://x.com/zebriez" target="_blank">https://x.com/zebriez</a></p><p>• The playbook for building high-talent-density teams | Adam Ward, Head of Talent at Cursor: <a href="https://www.lennysnewsletter.com/p/the-playbook-for-building-high-talent" target="_blank">https://www.lennysnewsletter.com/p/the-playbook-for-building-high-talent</a></p><p>• Building product at Stripe: craft, metrics, and customer obsession | Jeff Weinstein (Product lead): <a href="https://www.lennysnewsletter.com/p/building-product-at-stripe-jeff-weinstein" target="_blank">https://www.lennysnewsletter.com/p/building-product-at-stripe-jeff-weinstein</a></p><p>• Sutter Hill Ventures: <a href="https://shv.com" target="_blank">https://shv.com</a></p><p>• Snowflake: <a href="https://www.snowflake.com" target="_blank">https://www.snowflake.com</a></p><p>• Mike Speiser on LinkedIn: <a href="https://www.linkedin.com/in/mikespeiser" target="_blank">https://www.linkedin.com/in/mikespeiser</a></p><p>• Footnotes and Tangents: <a href="https://footnotesandtangents.substack.com" target="_blank">https://footnotesandtangents.substack.com</a></p><p>• The Power Broker Book Club: <a href="https://www.robertcaro.org/copy-of-six-books-six-ny-times-book" target="_blank">https://www.robertcaro.org/copy-of-six-books-six-ny-times-book</a></p><p>• <em>The Odyssey</em>: <a href="https://www.imdb.com/title/tt33764258" target="_blank">https://www.imdb.com/title/tt33764258</a></p><p>• <em>Rashomon</em>: <a href="https://www.imdb.com/title/tt0042876" target="_blank">https://www.imdb.com/title/tt0042876</a></p><p>• Akira Kurosawa: <a href="https://en.wikipedia.org/wiki/Akira_Kurosawa" target="_blank">https://en.wikipedia.org/wiki/Akira_Kurosawa</a></p><p>• Kevin Kwok on LinkedIn: <a href="https://www.linkedin.com/in/kevinakwok" target="_blank">https://www.linkedin.com/in/kevinakwok</a></p><p>• The Work You Do, the Person You Are: <a href="https://www.newyorker.com/magazine/2017/06/05/toni-morrison-the-work-you-do-the-person-you-are" target="_blank">https://www.newyorker.com/magazine/2017/06/05/toni-morrison-the-work-you-do-the-person-you-are</a></p><p>• Ari Weinstein on X: <a href="https://x.com/AriX" target="_blank">https://x.com/AriX</a></p><p>• Sky: <a href="https://sky.app" target="_blank">https://sky.app</a></p><p>• Dylan Field live at Config: Intuition, simplicity, and the future of design: <a href="https://www.lennysnewsletter.com/p/dylan-field-live-at-config" target="_blank">https://www.lennysnewsletter.com/p/dylan-field-live-at-config</a></p><p>—</p><p><strong>Recommended books:</strong></p><p>• <em>Barbarian Days: A Surfing Life</em>: <a href="https://www.amazon.com/dp/0143109391" target="_blank">https://www.amazon.com/dp/0143109391</a></p><p>• <em>Anna Karenina</em>: <a href="https://www.amazon.com/Anna-Karenina-LEO-TOLSTOY/dp/8175993421" target="_blank">https://www.amazon.com/Anna-Karenina-LEO-TOLSTOY/dp/8175993421</a></p><p>• <em>The Power Broker</em>: <a href="https://www.amazon.com/dp/0394720245" target="_blank">https://www.amazon.com/dp/0394720245</a></p><p>• <em>War and Peace</em>: <a href="https://www.amazon.com/War-Peace-Leo-Tolstoy/dp/8175992832" target="_blank">https://www.amazon.com/War-Peace-Leo-Tolstoy/dp/8175992832</a></p><p>• <em>Wolf Hall</em>: <a href="https://www.amazon.com/dp/0312429983" target="_blank">https://www.amazon.com/dp/0312429983</a></p><p>—</p><p>Production and marketing by <a href="https://penname.co/" target="_blank">https://penname.co/</a>. For inquiries about sponsoring the podcast, email <a href="mailto:[email protected]" target="_blank">[email protected]</a>.</p><p>—</p><p><em>Lenny may be an investor in the companies discussed.</em></p> <br /><br />To hear more, visit <a href="https://www.lennysnewsletter.com?utm_medium=podcast&utm_campaign=show-notes-no-free-preview-language">www.lennysnewsletter.com</a>
Key Insights
- Building for where models are now, where they will be in a year, or where they might be in one month are all equally wrong—the optimal window is two to three months, requiring close coordination with research teams on their roadmaps.
- At OpenAI, there is no treasure trove of secret strategy; instead, the company operates as founders-led where everyone in their domain acts as a founder with minimal top-down direction, creating a thin distance between product teams and market feedback.
- The shift from Stripe to OpenAI changed Seshan's shareable artifacts from long reasoning documents to mocks, prototypes, and experimental results because long documents no longer reliably signal rigorous thinking in an AI era.
- When all workers have access to the same AI tools, competitive advantage shifts entirely to human factors: ambition, taste, opinionation, and the ability to articulate a compelling vision for what the product should be.
- The most effective people using AI don't simply automate rote tasks but expand their range of possible capabilities, functioning like polymaths who can design, build, and reason about complex problems previously beyond their individual reach.
- A critical PM responsibility in the AI era is elevating others' ambitions by reminding teams what's actually possible and asking 'couldn't we try this faster or at 10x bigger scale?' rather than accepting initial estimates.
- Writing as thinking—the act of outlining, writing prose, and iterating to clarify ideas—should never be automated, but writing as reporting (summaries, status updates, translations between formats) should be automated as much as possible.
- Coding and knowledge work require fundamentally different product approaches: coding is output-oriented and verifiable through tests, while knowledge work requires transparency in reasoning, inputs, citations, and process to validate correctness.
- The current toggles between ChatGPT, Codex, and work mode are transitional states; the north star is a system where users describe their task and the interface automatically selects the right model and UI without explicit user choice.
- Product-marketing fit—the narrative, positioning, and pitch—can and should precede product-market fit; testing the story at enterprise by pitching to 100 people and refining the narrative is as important as building the product itself.
- The multiplayer future of work involves steering agents collectively as a team rather than working one-on-one with individual agents, requiring new interfaces and collaboration models between human teammates and their AI agents.
- Practical, prosaic infrastructure elements—data access, cloud reliability, agent integration with third-party systems—are as critical to effective agents as model intelligence itself for achieving real-world utility.
Topics
Transcript
If you think about the first era of AI products as chat, the second era of these products working with agents, that third era that might come soon is how do you work with a persistent co-worker who is able to get things done with you? There's this idea of the overhang of what AI is capable of and what we're actually doing with it. So hard to understand what is going to emerge in the future. You fail if you build for where the models are future. You fail if you build for where the models are now. You fail if you build for where you think the models will be in a year. Both outcomes are equally wrong.…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Lenny's Podcast: Product | Career | Growth
OpenAI’s Head of ChatGPT: We’re entering a new era of AI (again) | Tibo Sottiaux
Thibaut Sottiaux, Head of ChatGPT at OpenAI, discusses the launch of Dots (AI agents), the future of work with AI, and how builders should prepare for a world where most internet actions are performed by agents. He emphasizes the importance of safety, community-driven development, and the evolution from configuration-heavy systems to naturally learning AI assistants.
The grief, loneliness, and burnout sweeping through the tech industry right now | Molly Graham
Molly Graham revisits her famous 'give away your Legos' career advice in light of AI disruption, finding that while the core message about embracing change remains valid, the advice requires significant caveats when applied to delegating work to AI rather than humans. She argues that some work shouldn't be outsourced to AI, emphasizes the importance of acknowledging grief and emotions around change, and reframes AI as an intern rather than a superior intelligence.
90 minutes of unfiltered product advice from Snap and Discord’s product chief | Peter Sellis
Peter Sellis, former head of product at Snap and Discord, discusses building high-performing product organizations by organizing teams like ideological structures with clear decision-making authority, argues that most product managers are mathematically bad due to power-law distributions in skill, and shares lessons on focusing growth on core product value rather than expansion.
How we built Grok Bot in a month | Roman Ugarte (SpaceXAI)
Roman Ugarte, product lead at Grokbot (SpaceX AI), discusses how they built and launched a groundbreaking AI agent product in just one month by making two critical early decisions: running bots in the cloud with their own computers, and focusing ruthlessly on simplicity over features. The conversation covers their go-to-market strategy, the importance of manual onboarding, and how maintaining startup culture enables rapid iteration in a highly competitive AI market.
Why companies are becoming a series of loops | Anish Acharya (a16z)
Anish Acharya, a16z General Partner and former founder, discusses how AI is reshaping company organization through 'loops' (autonomous agent workflows), argues the permanent underclass fear is overblown, and contends that the real consumer opportunity lies in AI applications that improve happiness and human connection rather than just productivity.