Strug City is a fully autonomous agentic product and engineering organization, led by one human and built in public. Three products in production. One platform underneath.
The Story
Strug City is what happens when one operator decides to prove that the right AI team can build, run, and grow a real technology company. Led by Ryan Knollmaier — Founder & Chief Human Officer, with Sabine as Chief of Staff and an eight-agent Strug Works Virtual Engineering team — Strug City has shipped three products to production with three more in development.
It is an experiment, a company, and a working answer to one question: what can one human and the right AI team actually do?
Five acts
Project
A repo with no users
Experiment
What if agents ran the org?
Means
The org became the platform
Company
Products in production
Market
Building toward it
The Work
Each product solves an under-served need state. Each is built on Anti-Strug, our agentic operating system.
Personal AI super agent with memory, email intelligence, autonomous skill acquisition, and agentic shopping. Dual-context morning briefings with work/personal separation. FastAPI + Next.js deployed on Railway.
The autonomous virtual engineering platform that powers Strug City. Named in the tradition of Lockheed’s Skunk Works — a small, elite team producing breakthroughs without the overhead of traditional development.
Young athletes get a single home for their entire amateur career — development, stats, recruiting, communication, identity. Organizations and coaches get the indispensable tools their teams need to compete. The successor to GameChanger, SportsEngine, and TeamSnap.
The agentic cognition platform powering every Strug City product. Memory, intelligence routing, and multi-agent orchestration — absorbed Memory Lab and evolved into the shared intelligence layer for the Strug City ecosystem.
The Platform
Anti-Strug provides persistent memory, intelligent routing between agents, and multi-agent orchestration. Every Strug City product is a client. The platform improves every time any of them ships.
Strug City runs on Strug Works (built on Anti-Strug). This site is the first end-to-end production engagement. See the receipts →
MCP — Anti-Strug speaks open standards. Claude Code, Cursor, and any MCP-compatible agent can connect and interoperate.
The Thesis
We don't build features looking for a market. We identify under-served need states and build the products that solve them.
Every product on this site started with an honest answer to one question: who is the actual person being failed by the current options, and what would actually serve them?
Read the product thesis →Building in Public
Shipped the service layer for Unified Lists (PR #288): 352-line ListService with create, read, update, archive operations enforcing user isolation. Three TDD audit rounds caught 25 findings—security issues (missing RLS restore), type-safety degradations (ValidationInfo → object), and logic bugs (empty update → false 404). Completes the three-part infrastructure: migration → models → service.
Extended ListService with 4 item operations (add_items, update_items, delete_item) bringing the service to 646 lines. Three TDD audit rounds caught 14 findings: missing RLS, partial-failure tracking, reminder cascade ordering, atomic RPC for position race prevention. Ships with 1,126 lines of tests (9 CRUD + 12 audit).
Added name resolution to ListService with exact match, substring match, and NotFound outcomes. Returns Found (exact or single substring), Ambiguous (multiple substring matches), or NotFound with suggestions. 77-line method with 282 lines of TDD tests (T-21 through T-27) covering all resolution paths. TDD audit caught 5 findings: error handling gap, empty update validation, batch update invariant, type safety, dead code cleanup.
For the curious
A production AI Chief of Staff, live and working in the wild.
For the technical
Persistent memory, intelligent routing, multi-agent orchestration.
For the connected
Partnerships, collaborations, press — Ryan reads every email.