Lauren Tan
Software engineer on the SpaceX AI GrokBot team, known on platform X by the handle Poteto. A practitioner in the AI Agentic Coding domain, Lauren Tan is recognized for large-scale AI agent engineering methodologies and contributions to the discipline of building and operating AI agents at scale.
Timeline (last 90 days)
Lauren Tan's P-Stack deliberately includes no planning-type skills, with the philosophy that the best spec is the code itself, placing the emphasis on after-the-fact investigation, verification, and multimodal review.
Lauren believes verification is the most important skill in the AI programming toolbox, and that without verification an agent cannot form a true loop.
The autonomous merging approach is highly dependent on the verifiability of the domain, and there is currently no answer for scenarios that are difficult to verify programmatically and where PRs are one-way gates (irreversible data loss), such as certain medical, legal, and financial applications.
Lauren believes domain expertise is more important in the AI era, because the bottleneck has shifted from the agent to clearly expressing intent.
Lauren Tan (online alias Potato) is a former Meta React team member and currently an engineer at SpaceX AI (the Cursor team).
Lauren Tan submitted 2,500 PRs to the production environment within one month.
Lauren uses a cloud-based agent named Benny, which takes over all bug reports, automatically launches a desktop environment in the cloud, and uses control skills to reproduce the issues.
Lauren has directly disabled useEffect and code comments in CI, on the grounds that in 99% of cases, comments written by agents only document irrelevant historical context.
Cursor engineer Lauren Tan joined Cursor from Meta's React team five months ago.
Lauren Tan has already had an AI agent automatically merge PRs into the main branch on her behalf.
7 more timeline events
All Facts (16)
Lauren Tan's P-Stack deliberately includes no planning-type skills, with the philosophy that the best spec is the code itself, placing the emphasis on after-the-fact investigation, verification, and multimodal review.
50%UnverifiedLauren believes domain expertise is more important in the AI era, because the bottleneck has shifted from the agent to clearly expressing intent.
50%UnverifiedThe autonomous merging approach is highly dependent on the verifiability of the domain, and there is currently no answer for scenarios that are difficult to verify programmatically and where PRs are one-way gates (irreversible data loss), such as certain medical, legal, and financial applications.
50%UnverifiedLauren Tan (online alias Potato) is a former Meta React team member and currently an engineer at SpaceX AI (the Cursor team).
50%UnverifiedLauren Tan submitted 2,500 PRs to the production environment within one month.
50%UnverifiedLauren believes verification is the most important skill in the AI programming toolbox, and that without verification an agent cannot form a true loop.
50%UnverifiedLauren uses a cloud-based agent named Benny, which takes over all bug reports, automatically launches a desktop environment in the cloud, and uses control skills to reproduce the issues.
50%UnverifiedLauren believes one should invest as much as possible in mechanisms that can be hard-enforced, which is also why Rust is gaining popularity again — if the code compiles, it will most likely run correctly.
50%UnverifiedCursor engineer Lauren Tan joined Cursor from Meta's React team five months ago.
50%UnverifiedLauren Tan has already had an AI agent automatically merge PRs into the main branch on her behalf.
50%UnverifiedLauren believes the most important capability when collaborating with agents is verification — having the agent actually run code, capture CPU traces, heap snapshots, or open an iOS simulator to verify functionality.
50%UnverifiedIn greenfield projects, agents organize code along the path of least resistance, and without guardrails, code structure rapidly degrades as features are added — what Lauren calls the organic architecture risk.
50%UnverifiedLauren has directly disabled useEffect and code comments in CI, on the grounds that in 99% of cases, comments written by agents only document irrelevant historical context.
50%UnverifiedDeveloper Lauren Tan (online alias Poteto) merged approximately 1,000 Pull Requests in July
50%UnverifiedLauren encapsulated fixes for mistakes the Agent repeatedly made into reusable skills and automated code-check rules, so that all subsequent work is validated against them
50%UnverifiedLauren Tan merged 2,462 Pull Requests in August
50%