Where AI products go next: voice, agents, and self-driving software | Tara Sesha and Nan Yu (OpenAI)
OpenAI product leaders Tara Seshan and Nan Yu join Lenny's Podcast to talk about building AI products fast, designing agents, and what product strategy looks like heading into 2027.
The gist
-
Ship imperfect but working features quickly so real user feedback guides decisions, rather than over-planning before launch.
-
Design agents around a hierarchical structure, like a chief of staff managing others, so users are not juggling dozens of independent agents.
-
Plan only 2 to 3 months ahead in fast-moving AI markets, since predictions beyond that horizon are almost always wrong.
-
Build a fallback layer, such as Computer Use, so users can finish tasks even when pre-built integrations fail.
Claire Vo uses it
"I just love computer use now. Like don't even give me an MCP. I want to point computer use at your UI."
▶ Play at 13:022 creators recommend it, including Yana Welinder.
See all on the Computer Use pageWhat they recommend · 3 picks
Codex
AI tools
"My beloved Codex docked on my computer that I talk to all day."
Computer Use
AI tools
"I just love computer use now. Like don't even give me an MCP. I want to point computer use at your UI."
Grok
AI tools
"My 40 Grok bots that are load-bearing across different parts of my business."
Also mentioned
◎
Related episodes
Episodes covering similar topics
More from Lenny's Podcast