The Platform
Six components, one operating system. Every layer is governed, audited, and evidence-backed.
AI Fabric
The single entry point for every AI task. It applies policy, routes to the right model, and records an append-only audit trail for each execution.
VERIFIED
AI Policy
Classifies each task (coding, vision, offline, auto) and decides which route and model class may handle it. Policy is explicit, not implicit.
VERIFIED
OmniRoute
The model router: 874 models across 19 provider namespaces, with per-route health checks and fallback behaviour. No direct provider calls from applications.
VERIFIED
Control Centre
The operator console: service health, company portfolio, opportunities, approvals, and the AI execution audit — with honest NOT MEASURED states.
VERIFIED
Paperclip
The agent orchestration layer: missions, goals, issues, and heartbeats for the AI workforce. Agents execute with defined roles and chain of command.
VERIFIED
Chatwoot
The canonical conversation inbox. Email, web leads, and agent activity all land in one place — no second CRM, no duplicated inboxes.
VERIFIED
Runtime evidence
| Capability | Status | Evidence |
|---|---|---|
| AI execution | VERIFIED | 41 executions, 32 SUCCESS, 9 DEGRADED (audit IDs recorded) |
| Skills execution | VERIFIED | 19 skills executed at runtime, all SUCCESS |
| Agent runtime | VERIFIED | 7 agents with runtime evidence |
| Model routing | VERIFIED | 4 executable providers; 3 auto routes DEGRADED (documented) |
| Regression tests | VERIFIED | 240 PASS across 14 test files |
| Company isolation | VERIFIED | 12/12 isolation checks passing |
| Secret scanning | VERIFIED | gitleaks: 0 leaks found |
Evidence captured 2026-09-04. Full details in the public insights report.
Want the full architecture?
Book a discovery call for a live walkthrough of the platform and its audit trail.
Book a Discovery