Ninebrains, Apache-2.0 · macOS, Windows, Linux
Run Claude Code and Codex agents side by side, each in its own git worktree. A second, independent run has to prove the work before it counts as done.
- 4
- 3
- <5s
We build the instruments agents run on.
Advance Labs is an independent lab building open-source software for the agent era (Ninebrains, runs-on.dev and the AEO Toolkit) and selling the expertise that built them as monthly AEO retainers and build sprints.
Ship the product first. Then sell what it taught us.
Ninebrains, runs-on.dev and the AEO Toolkit. Open source at the core, in use, and the reason anyone should believe the rest.
AEO audits and build sprints, sold at a fixed price by the people who built the tools behind them.
Drone autonomy in simulation and quantum-hybrid feasibility work. Published with the code, so the claims can be checked.
Three products. Two fully open source, one open core.
runs-on.dev, AGPL-3.0 · One name per GitHub account
Free subdomains for developers. Sign in with GitHub, claim yourname.runs-on.dev, and the record is one public file anyone can read.
- $227,000
- $0
AEO Toolkit, Open core · Apache-2.0 engine
Tools that measure whether answer engines can find, parse and cite a site. Free in the browser, in Chrome, and over MCP, on an open-source scoring engine.
- 58
- 40
- 32
Numbers you can check, work you can open.
23
- 6
- 275
- 228
Kuzyn Builds → BuildCode
Designed and built the marketing site for a London, Ontario custom-home builder, then set up its local AI-search visibility. The relationship grew into BuildCode: plain-English Ontario building-code lookup for the trades.
“Introduced me to Claude, which has revolutionized my business.”
Smart Fix London
Cleaning up how Google and AI assistants see a London, Ontario repair shop: one name, address and phone across every listing, one clean homepage, full LocalBusiness structured data, and service pages for the repairs AI answers were missing.
Converg3nce
An AI-visibility venture for commerce brands: it read a brand's web presence the way an AI shopping agent does, scored how agent-ready it was, and ranked the fixes. Wound down in 2026; the site is still up.
The expertise that built them, by the month or by the sprint.
Two ways to buy the work, both with the price stated before anything starts.
AI Search Visibility Retainer
Monthly answer-engine optimization. Your buying questions re-run in ChatGPT, Perplexity and Google every month, the fixes that keep you out of the answers shipped, and a one-page report. Managed $500 a month, Growth $1,000 a month.
Fixed-Scope Build Sprint
A scoped chunk of engineering for a written scope and a fixed price. AI integrations and automations, or conversion-focused Next.js/React web builds.
The long bets, published with the code.
Research runs in the open so the claims can be checked rather than taken on trust.
Urban drone autonomy
A multirotor that plans its own mission through real cities imported from OpenStreetMap, with a separate safety layer that can overrule any move. Sim-first, 275 tests and a first PX4 SITL flight, Apache-2.0.
Quantum-hybrid research
Feasibility studies where quantum meets classical computing, including whether quantum hardware can speed up LLM training. Every claim is tagged by the strength of its evidence; the 228-test emulator is public.
Tell us what you are building.
What is changing, where the workflow breaks, and what a useful outcome would look like.
Already know what you need? See the monthly AEO retainer or the fixed-scope build sprint.