Lenny's Podcast: Product | Career | Growth

How we built Grok Bot in a month | Roman Ugarte (SpaceXAI)

September 8, 2026

AI Summary

5 min read

In early 2025, a small team inside the AI company formerly known as Cursor (now part of SpaceXAI) went into a separate part of the office, created private Slack channels, and spent a month building a knowledge-work product from scratch. The result was Grok Bot. Three weeks after its public launch, the team was hosting standing-room-only meetups. Roman Ugarte, who led product for the effort, explained the two early, non-obvious decisions that made the product work — and the operating principles that let a small team ship something that felt categorically different from everything else on the market.

The two decisions that broke the mold

Ugarte identified two architectural choices that, in hindsight, defined Grok Bot's success. First, the team decided that every bot should run entirely in the cloud, not on the user's local machine. This meant a user could kick off a task from their phone, close their laptop, and the bot would keep working. "You should never have to think about local and cloud and where are these workflows running," Ugarte said. "Does my computer have to be awake? If I kick it off from my phone, does it need to be tethered to my computer back at home?" The cloud decision made the bot a persistent, always-available entity — a colleague with its own runtime, not a session tethered to a device.

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

Never miss an episode of Lenny's Podcast: Product | Career | Growth

Get every new episode summarized in your inbox — free, ~5 minutes to read.

No spam. Unsubscribe anytime.

What you'll learn

1 (00:00) **Introduction & The "100% of the Job" Thesis** - Lenny introduces Roman Ugarte, employee #15 at Cursor and product lead for Grok Bot, which Lenny calls the "hottest AI product in the world."
2 (04:12) **Origin Story: The "Cave" Month** - A small, isolated team built the initial prototype from scratch in one month.
3 (06:05) **Internal Beta & The "All-Hands" Moment** - The internal rollout was a critical pressure test that validated the product's value.
4 (09:07) **Key Decision #1: Starting Fresh vs. Building Inside Cursor** - The controversial choice to build a new product rather than a feature inside Cursor was critical.
5 (11:27) **Manual Onboarding: The "Painful" Feedback Loop** - The team manually onboarded 200-300 early users over two weeks to learn rapidly.
6 (14:32) **Key Decision #2: Hiding the Internal Mechanics** - Grok Bot deliberately hides the agent's internal tool calls and thought process.
7 (18:41) **The Three Weeks Before Launch: "Unshipping" and "Hill Climbing"** - The team aggressively simplified the product and focused on reliability.

+ Full timestamped outline available in the app

Show Notes

Roman Ugarte helped incubate and build Grok Bot, the popular new knowledge-work agent from SpaceXAI. A small, isolated team took it from first line of code to a working internal product in four weeks, and to a hugely successful public launch just three weeks later. Before Grok Bot, Roman led Growth at Cursor, where he helped scale the company from 15 people to over 1,000 before its acquisition by SpaceX.

In our in-depth conversation, we discuss:

1. The origin story of Grok Bot

2. The key decision to build it from scratch instead of adding it to Cursor

3. Why the team personally onboarded nearly 300 of its first users

4. The two early product decisions that made Grok Bot so successful

5. Their “colleague-pilled” product philosophy

6. Roman’s advice on moats, and what has allowed Cursor to keep winning in the most competitive market in the world

—

Brought to you by:

WorkOS—Make your app enterprise-ready, with SSO, SCIM, RBAC, and more

Mercury—Radically different banking, now with Command

—