Roman Ugarte: How to Onboard AI Agents Like Employees
Roman Ugarte shares his 4-step method for onboarding Grok Bot agents to audit your work, triage bugs, and build automated daily digests.
How we built Grok Bot in a month | Roman Ugarte (SpaceXAI)
With Lenny Rachitsky, Roman Ugarte · Sunday, September 13, 2026
Roman Ugarte, product lead at SpaceXAI and former Head of Growth at Cursor, details how a small, isolated team built and launched Grok Bot in just seven weeks. He shares the product decisions behind Grok Bot's breakout success—including providing each bot its own dedicated cloud VM and treating AI like human colleagues rather than feature-dense software. Roman also discusses the culture of unshipping product scaffolding, why startup moats are discovered rather than planned, and the team's philosophy on manual onboarding and rapid execution.
Key takeaways
- Roman Ugarte and a small SpaceXAI team built and launched Grok Bot in seven weeks, giving each bot its own cloud virtual machine and treating it like a human colleague. Read more →
- The Grok Bot core team built and shipped the product in seven weeks, manually onboarding 200 to 300 users one by one. Read more →
- Cursor and SpaceXAI outpace better-funded foundation model labs by discovering moats through rapid shipping rather than upfront strategic planning decks. Read more →
- The Grok Bot team built and shipped the product in seven weeks, spending the final two weeks actively removing experimental UI and feature panels. Read more →
4 articles from this episode
Why Grok Bot Manually Onboarded Its First 300 Users
Roman Ugarte explains why SpaceXAI manually onboarded 300 early Grok Bot users to spot bugs and discover emergent AI workflows.
Why Deleting Product Code Beats AI Moat Diagrams
Roman Ugarte explains how Cursor and Grok Bot won by rebuilding every six months and deleting scaffolding as models improved.
Why Grok Bot Unshipped UI for 100% Task Delegation
Roman Ugarte explains why Grok Bot deleted UI buttons and settings to reach 100% AI task delegation instead of stopping at 90%.
More Lenny's Podcast episodes
- An 86-minute guide to finding PMF | Rob Snyder (Author, The Power of Pull)Oct 25 · 4 articles
- A mind-expanding conversation on how models actually work | Emmanuel Ameisen (Anthropic)Oct 18 · 6 articles
- Why every company now needs to think and operate like a lab team | Josh Woodward (VP Google Labs)Oct 11 · 3 articles
- Where AI products go next: voice, agents, and self-driving software | Tara Sesha and Nan Yu (OpenAI)Oct 4 · 5 articles