Garry Tan System Map

Module Build Plan

The completion plan, evidence standard, and module inventory.

gstack = methodgbrain = continuitygbrain-evals = proofYC = network

Module Build Plan

Objective

Complete a public, source-backed reverse engineering map of Garry's AI software factory. Each module should be useful on its own, but together they should read like a product spec, GTM teardown, JTBD map, and contribution strategy for the whole system.

Definition of Done

Each module is complete when it has:

Module Inventory

ModuleStatusCore QuestionPrimary Evidence
01 gstack teardownComplete v1What product does the 52-skill workflow actually define?gstack README, SKILL.md files, docs
02 gbrain runtime mapComplete v1How does persistent memory work as a product and data system?schema.sql, core types, README, MCP docs
03 gbrain skillpack JTBDComplete v1What work does the 43-skill brain workflow automate?gbrain/skills SKILL.md files, README
04 funnel and CTA teardownComplete v1How does OSS attention turn into YC hiring and founder demand?gstack README, YC software/apply pages
05 competitive mapComplete v1Where does this system sit vs coding agents, memory tools, and eval repos?README claims, repo capabilities, market categories
06 contribution mapComplete v1Where can Ren create credible, mergeable value?open issue/PR themes, repo health, duplicate-risk scan
07 gbrain-evals proof layerComplete v1How does the system make memory claims falsifiable?gbrain-evals README, benchmark reports, comparison docs, runner layout

Build Order

  1. Build the internal source-backed docs first.
  2. Mirror each module onto the public Pages site.
  3. Keep the home page as the executive map, and make /modules/ the working index.
  4. Add deeper module revisions as new evidence comes in.

Completion Standard

This should not stop at "interesting analysis." The final system should let us answer: