OpenAI's product lead Tara Seshan has outlined the company's vision for the next era of AI: persistent coworkers that collaborate with users over time. This follows the chat and agent eras, and is designed to work with models that are 2-3 months ahead of current capabilities.
TL;DR
- OpenAI's third era of AI focuses on persistent coworkers that collaborate with users over time.
- The company is building for models that are 2-3 months ahead of current capabilities.
- OpenAI's approach favors speed and empirical testing over academic strategy.
What happened
OpenAI's product lead Tara Seshan has outlined the company's vision for the next era of AI: persistent coworkers that collaborate with users over time. This follows the chat and agent eras, and is designed to work with models that are 2-3 months ahead of current capabilities.
Seshan, who leads Codex and ChatGPT Work, the company's fastest growing product for knowledge workers, shared her insights on Lenny's Podcast. She emphasized the importance of building for models that are 2-3 months ahead, as building for current or far-future models would lead to failure.
Seshan described the shift in product management, noting that in the fast-moving AI market, being prolific and empirical beats being academic or theoretical. She emphasized the importance of sharp hypothesis definition, fast tests with users, and feeding results back into the loop.
Why it matters
If work becomes steering fleets of persistent agents, the interface winners are those that solve collaboration, permissions, and provenance across many agents at once. Incumbents with distribution like ChatGPT can push agents to a billion users in one move, while developer-led tools must win on depth and trust with engineers.
The 2-3 month horizon also reorders roadmap bets, favoring teams that can ship, measure, and rewrite quickly over those that bet on a 12-month plan. This approach favors startups that can iterate quickly and respond to user feedback, rather than those that rely on long-term planning.
OpenAI's approach to building persistent AI coworkers could set the standard for the industry, as the company has the resources and distribution to push these capabilities to a wide audience. However, the success of this approach will depend on the company's ability to solve the challenges of collaboration, permissions, and provenance.
Key facts
- OpenAI's third era of AI focuses on persistent coworkers that collaborate with users over time.
- The company is building for models that are 2-3 months ahead of current capabilities.
- OpenAI's approach favors speed and empirical testing over academic strategy.
- OpenAI raised $100B in 2026 at a post-money valuation of $850B, according to StartupHub.ai.
- Seshan spent six years at Stripe as one of its first five product managers and was repeatedly ranked in Stripe's top three.
- Seshan later led product at Watershed, founded a company, was a Thiel Fellow, and was a Lenny's Newsletter Fellow.
- OpenAI's Codex and ChatGPT Work are the company's fastest growing products for knowledge workers.
- OpenAI's approach to building persistent AI coworkers could set the standard for the industry.
Context
OpenAI is a leading AI company known for its work on large language models, including ChatGPT and Codex. The company has raised significant funding and has a post-money valuation of $850B, according to StartupHub.ai.
The AI industry is rapidly evolving, with companies racing to develop the next generation of AI capabilities. The shift to persistent AI coworkers represents a significant evolution in the way AI is used, moving from chat and agent-based interactions to collaborative, long-term relationships.
The success of this approach will depend on the ability of companies to solve the challenges of collaboration, permissions, and provenance. This will require not only advances in AI capabilities, but also in the infrastructure and tools that support these capabilities.