Watch the replay when it lands, explore the actual Lenny's Podcast package from the session, then use the same method to map one workflow of your own.
One connected record of what the team recommended, built, checked, and handed back. Explore freely or let the tour move you through the five views.
It gives Sara the sharpest product-builder angle and a clear continuation of two recent conversations.
Sara DavisonSara Davison · Co-Founder and Strategy Lead at AI Build Lab
The episode: Sara explains why agent workflows break at the handoff, how to decide what agents should own, and where human judgment creates the most leverage. She walks through a practical three-part model product teams can use to define ownership, checkpoints, and clear stop conditions.
A simple way to choose the first workflow, place human judgment, and stop weak output before it compounds.
A live product-team walkthrough that turns an abstract agent discussion into decisions listeners can see.
A tight episode spine, one worked example, and three clear audience takeaways.
Sara would show how to design that human layer on purpose.
Sara's angle keeps human judgment from becoming the next bottleneck.
Hi Lenny's team,
Boy, do I have a speaker for you.
If you haven't heard of Sara Davison, you're about to. She co-founded AI Build Lab, which has taught more than 2,000 students across 15+ countries.
Sara has a gift for taking complicated AI systems and making them useful. She can show Lenny's audience how to decide what an agent should own, where a person needs to step in, and when to stop a workflow before bad output compounds.
You should have her on. I put together a short packet that lays out the episode.
If it clicks, I'll make the intro.
Hunter
AI Build Lab
The useful pattern is not “more agents.” It is a visible chain where every role hands forward an artifact, evidence travels with the decision, and external action stops with a person.
Find options and keep the evidence.
Compare tradeoffs and choose honestly.
Turn the choice into usable artifacts.
Fix weak claims before handback.
Preparation never becomes outreach by accident.
Paste one prompt into Codex, Claude Code, or another capable coding agent. It first asks permission to read this lesson, walks you through the example, then turns one workflow of your own into something you can see and improve.
Want to keep it? At the end, your agent will tell you that you can ask it to save the exact tagged Workflow X-Ray skill. It must show the files, destination, and hash, then ask before writing.
No installation, account, or email gate is required.
Bring a workflow you care about. Learn how to design the roles, proof, handoffs, and human boundaries that make agent work useful.