Core Platform
Process management for the AI era
A platform where every operational process has a human-readable form, a named owner, full version history, and an MCP interface. Accessible to people and AI agents simultaneously.
Platform capabilities
Everything a process needs to stay alive
Markdown + YAML frontmatter
Every process is a human-readable Markdown document with structured YAML frontmatter, following the Agent Skills spec. Editable by anyone, parseable by any AI.
Immutable version history
Every save creates a new immutable version. View diffs, roll back to any point, and see which version an AI client consumed on any given call.
Named ownership
Every department and process has a named owner, a real person accountable for keeping it accurate. Ownership is the accountability anchor.
MCP delivery layer
An MCP server exposes your processes to any compatible AI client. Three tools: read_process, list_processes, and update_process for AI-assisted authoring.
Usage analytics
Every MCP call writes a structured event. See adoption rates, per-client breakdowns, staleness pressure, and skill health scores per process.
Staleness alerts
Configurable rules trigger email or webhook notifications when a process hasn't been reviewed in N days, or has been called many times without an update.
Org hierarchy
Structure that mirrors how your org works
Koinoflow organizes processes around a four-level hierarchy that maps naturally to how real companies operate. Ownership cascades down: every department and every process has a named accountable person.
Workspace
Acme Corp · Top-level org unit
Team
Engineering · Division or business unit
Department
Frontend · Functional area with an owner
Process
Deploy to production · Owned procedure, the atomic unit
// Every MCP call writes a usage event:
{
"process_id": "proc_4f8a2b",
"process_slug": "engineering/deploy-to-production",
"version": "2.4.1",
"client_id": "cursor-workspace-abc",
"client_type": "cursor",
"called_at": "2026-04-11T09:12:33Z"
}
// Aggregated analytics:
{
"process": "Deploy to production",
"calls_30d": 847,
"unique_clients": 12,
"last_updated": "2026-01-15", // 86 days ago
"health_score": 34,
"alert": "staleness: review overdue"
}Observability
Turn AI usage into actionable insight
Every MCP call writes a structured usage event. The analytics dashboard turns that raw log into adoption rates, staleness pressure, per-client breakdowns, and skill health scores.
- "Your onboarding process was called 847 times last month"
- "It hasn't been updated in 90 days. Here are the 3 AI clients still on v1.2."
- "Your Deploy process has a health score of 34. It's stale and unowned."
Included in all plans.
Give your AI agents a single source of truth
Start with a free workspace. MCP server running in 30 minutes.