Overview
One GPU lane, one CPU lane, cloud on call. Hardware refreshes every few seconds.
Operations
Loaded models
Lane caps
GPU lane4 concurrent
CPU lane2 concurrent
Cloud laneClaude XO · Codex escalation
What the numbers say
- The orchestrator does not move the score: the same platoon scored 24/52 under our harness and 24/52 under n8n.
- Local Qwen 27B fails at discovery. Every local arm scored 0 on the spreadsheet, Plane-issue and résumé tasks.
- The cascade formation won the 3D tournament with 29/45 in 18 calls. Debate cost 50 calls for 24/45.
Benchmarks
Every unit is ranked by measured results on checkable tests.
TAC score by arm
Cloud modelLocal formation
ArmTAC score (0–100)PointsScore
Official TheAgentCompany graders · 10 tasks · 52 checkpoints. Partial arms are still running.
Points per task
0Full marks
Formation tournament · 3D modelling
Score out of 45 · same tasks for every structureFormationScore/45CallsMinutes
Unit qualification tests
Our runPublished reference
Market
305 products mapped and checked against live pages on 13 Oct 2026.
Rainbow Regiment is the organisation layer for a local-first AI company. It turns your own GPU box and a few cloud models into ranked, benchmarked units you command from one chat.
305
products mapped
151
run fully local on Windows, no account
7
"AI company" products, none lane-aware
±0
TAC points from changing orchestrator
Where we fit
| Layer | Market state | Main players | Our move |
|---|---|---|---|
| Organisation | Thin and weak | Paperclip, Multica, MetaGPT, ChatDev, Composio, CrewAI AMP, Claude Managed Agents | This is the product |
| Memory | Vendor claims unreliable | Mem0, Letta, Zep, Cognee | Logbook from checker-verified runs |
| Worker harnesses | Strong, mostly open | OpenHands, Codex CLI, Claude Code, OpenCode, Goose | Plug in; main lever on scores |
| Agent frameworks | Free | Pydantic AI, MS Agent Framework, LangGraph | Use as parts |
| Durable execution | Free, proven here | DBOS, Hatchet, Temporal, Conductor | Build on DBOS |
| Triggers | Commodity | Zapier, Make, n8n, Activepieces | Use one |
Lane drill · hands-on
8-job plan, GPU 4 / CPU 2 caps across two runs, killed mid-run| Tool | Layer | Caps across runs | Crash test | Verdict |
|---|
The gap nobody combines
- Hardware-lane awareness on a home box
- Cloud brains and local hands, measured as score per cloud dollar
- Units that earn rank on published, independent benchmarks
- One front door over many chats
- A findings register with a conflict ledger
- Local-first and open
Risks
Vendors add an org layerClaude Managed Agents, OpenAI
Local-model ceilingvalue must show in numbers
LM Studio licenceno hosted service
Open-core partsDBOS UI, Windmill, Kestra
Units
Each unit has one mission, a checker and a score. Rank thresholds are not set yet, so every unit is provisional.
| Role | Assigned | Kind | Mission | Benchmark | Rank |
|---|