Grok Bot: What Changes When You Can Build AI Teammates
Lauren Tan and Roman Ugarte discuss Grokbot, a platform for creating AI teammate agents with their own identities and tool access that can work autonomously. The conversation covers how AI agents are transforming software engineering and non-technical work, with Lauren showcasing her 4,000+ PRs/month workflow using specialized bots in Slack and describing the shift from execution-constrained to execution-abundant organizations.
Summary
Grokbot is a product that enables users to create AI teammates with distinct identities, functions, and tool access—essentially autonomous agents that can perform work alongside humans. Lauren and Roman explain that the product emerged from observing how AI coding transformed engineering workflows, leading them to build something that works for non-engineers too. The core principles are: organizing work around bots rather than chats or files, giving bots access to the same tools humans use, and making them always-on and proactive.
The reception has been surprisingly broad, with both highly AI-forward and skeptical employees finding value. The team has observed a trust-building pattern: users start with simple tasks, gradually expand bot capabilities, and eventually delegate full job responsibilities. Real-world deployments show marketing teams using bots to manage ad campaigns, support teams automating ticket triage, and small businesses using bots to handle functions they cannot afford to hire for.
Lauren Tan shares her personal workflow, which has evolved significantly. She uses multiple specialized bots (Potato being her primary engineering bot) primarily through Slack and X rather than the Grokbot app itself. Her process now involves creating bots with specific responsibilities, connecting them to tools and data sources, and letting them work collaboratively while she focuses on product direction and high-level decisions. She's merged 4,000+ PRs monthly by having her bots run code, verify their own work through empirical testing, and in some cases merge their own pull requests.
The discussion touches on P-Stack, Lauren's open-sourced engineering stack that teaches agents how to write code rigorously. The key insight is that rather than relying solely on costly token-intensive code review processes, engineers should invest in codebase infrastructure—static analysis, lint rules, architectural constraints—that encode quality standards and reduce what agents need to review. This inverted pyramid approach (heavy investment in codebase, lighter investment in AI review) is more token-efficient and produces better results.
Grokbot's multiplayer features, recently launched with team bots, allow domain experts to build and share bots across organizations while maintaining proper permissions and authentication. This creates a pattern of AI employees managed by their respective domain experts, similar to hiring a specialist for a particular business function.
The conversation explores how this abundance of execution capability is shifting organizational bottlenecks from engineering capacity to alignment and vision. With bots capable of rapidly building features, the constraint becomes defining what to build, maintaining consistent product philosophy, and exercising taste in what features ship. The team emphasizes intentional design and dog-fooding over shipping every possible feature.
Lauren's personal story illustrates accessibility to AI engineering expertise: she studied accounting and finance, taught herself to code through building a startup, shared knowledge on X and blogs, and was eventually recruited into framework communities. This led to roles at major tech companies and now Grokbot. The common thread is public communication of ideas and consistent iteration.
Key Insights
- Grokbot founding principle: organize work around bots as the operative unit, not chats or files, with bots having access to the same tools humans do and running continuously rather than waiting for user prompts.
- The trust-building pattern for AI agents follows a progression: users start with simple read-only tasks, verify results, gradually expand bot access and responsibility, then eventually delegate full job functions to agents.
- Lauren's engineering loop now operates primarily through Slack threads with specialized bots, eliminating the need to context-switch between chat and development environments, with features built entirely in public collaborative threads rather than private development.
- P-Stack's inverted pyramid approach prioritizes encoding quality standards into the codebase through static analysis and architectural constraints rather than expensive token-intensive code review, reducing both costs and improving code quality simultaneously.
- With abundant execution capability through AI agents, organizational bottlenecks shift from engineering capacity to alignment on vision, requiring teams to exercise taste, say no to features, and maintain high bars for what ships to users despite rapid development speed.
Topics
Transcript
[0:00] Grokbot is meant to be this product where you can create AI teammates. >> They have their own identity, their own telephone number, credit card, like basically almost anything uh a human can do, I think we want bots to do. I sort of see myself as the explorer, you know, like trying to figure out what will what does software engineering look like 6 months through a year from now. >> Welcome back to another episode of The Light Cone. Today we're joined by Lauren [0:32] Tan and Roman Ugarte, the team behind Grokbot. Roman and Lauren incubated the product and are on the core team that took it from the first line of code to launch in…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Y Combinator
We’ll Be Inferior to Our Grandchildren—and That’s Great
A discussion about scaling societies and the relative ratio of good versus bad behavior, with one participant expressing concern that future generations will be intellectually superior to current humans, similar to how humans are vastly superior to ants.
YC Debates: Should We Slow Down AI?
Jacob Coxin argues governments should substantially intervene to slow frontier AI development, citing rapid capability growth (superhuman AI researchers by 2027), alignment risks, and misaligned corporate incentives. Robin Hanson counters that typical technological adoption patterns suggest slower real-world impact, that regulation is dangerous without concrete problems, and that liability-based approaches are preferable to centralized global regulation.
How To Build The Future: Eric Glyman, Co-CEO of Ramp
Eric Glyman, co-CEO of Ramp, discusses how the company evolved from Parabus to build a $60 billion financial operations platform, and how Ramp is leveraging AI agents and internal tools like Inspect, Research, and Glass to transform organizational productivity and reshape the future of work.
Grok Bot: What Changes When You Can Build AI Teammates
Lauren Tan and Roman Ugarte discuss Grockbot, a product enabling users to create AI teammates with their own identities and capabilities. They share how AI agents are transforming software engineering and business operations, demonstrating workflows where engineers ship thousands of PRs monthly by leveraging specialized bots integrated into team communication tools.
Is AI Now Good Enough To Cure Cancer? | YC Paper Club
The YC AI for Bio Paper Club event explores how artificial intelligence is transforming cancer research and biological discovery through foundation models, personalized genomic analysis, and autonomous research systems. Speakers demonstrate that cancer is becoming a data problem solvable through AI agents, DNA language models like EVO2 are enabling efficient protein and genome design through few-shot prompting, and validated lab results show these approaches are moving from theory into clinical practice.