How Plait Works

From tribal knowledge to living infrastructure

Three layers. One living graph. Real-time answers for humans and AI agents — built to stay current because your systems depend on it.

Right Now

Your operating model lives everywhere and nowhere

Three systems that are supposed to hold the answer. None of them actually do.

HRIS / Workday
Reporting lines only
Shows who reports to whom. Nothing about decision rights, service ownership, or approval flows.
Not enough
Wiki / Confluence
Stale within weeks
Last updated six months ago. Half those people have left. Search returns twelve contradictory pages.
Out of date
Slack / Email / Heads
Invisible and fragile
Where the real source of truth actually lives. Not searchable. Not machine-readable. Leaves when people do.
Not transferable
Three broken sources, zero reliable answers Plait replaces all three with a single, living, machine-readable graph
Architecture

Built in three layers

From the graph at the core, to the integrations that keep it alive, to the experiences your team actually uses.

1
Knowledge Graph
The foundation — temporal, compliance-ready, always current

A structured graph database capturing your organisation's structure, accountability, capabilities, and workflows — with full temporal versioning so every historical state is preserved for compliance and audit.

24+ object types: teams, roles, services, decision authorities
Full temporal history — every change versioned and timestamped
Compliance-ready audit trail for every accountability change
Machine-readable at the schema level — no conversion needed
2
Integration Layer
REST API + MCP — other systems query Plait to function

Plait connects to the systems your teams rely on every day. When Plait is the source of truth for routing, approvals, and assignments, it stays current because broken integrations are immediately visible.

REST API for any system to query org authority
MCP interface for AI agents and copilots
Jira auto-assignment based on Plait ownership
Slack/Teams bot answering org queries in real time
3
Experience Layer
Org Canvas, search, Slack bot — zero training required

Clean, intuitive interfaces your team actually uses — without needing training. The Org Canvas gives visual clarity, full-text search finds any role or service instantly, and the Slack bot puts answers where questions are already being asked.

Org Canvas — visual operating model with live data
Universal search across roles, teams, services, decisions
Slack/Teams bot for instant query resolution
Role editor with guided accountability templates
In Practice

What happens when someone asks "who owns authentication?"

1
Question in Slack
Sarah asks in #engineering-general. Plait bot intercepts automatically.
2
Graph query
Plait identifies owner, decision rights, and approval chain in milliseconds.
3
Answer returned
Structured, sourced, and current — with last-updated timestamp.
4
Downstream actions
Jira auto-assigns. Approvals route through Plait. Everything works.
#engineering-general · 312 members
SL
Sarah Lin 9:47 AM
Hey team — who owns authentication? Need to loop in the right person for the new SSO implementation.
P
Plait 9:47 AM · instant
Here's what the operating model says about authentication:
🔐 Authentication · Service Record
Service ownerPlatform Engineering
Team leadMarco Rossi Tech Lead
Decision rightsMarco Rossi for architecture; CFO sign-off for spend >€50K
Approvals requirePlatform Lead + Security (policy changes)
Last updated3 days ago by Marco Rossi
SL
Sarah Lin 9:48 AM
Perfect, exactly what I needed. Thanks Plait! 🙌
AI Readiness

The missing layer for enterprise AI

AI agents are ready to work. But they can't route decisions, check authority, or escalate appropriately without a machine-readable operating model. That's Plait.

79%
of enterprises have deployed AI agents — but most are running blind on org authority
75%
cite unclear decision rights as the primary blocker for AI governance
Plait MCP · AI agent query
// AI agent queries Plait via MCP
plait.query("who approves infra spend >€50k?")
→ Response
approver: "Platform Lead"
secondary: "CFO sign-off required"
threshold: 50000
currency: "EUR"
plait.query("who can approve prod deployments?")
approver: "Tech Lead + Release Manager"
emergency: "CTO override available"
sla: 4 // hours
// No hallucinated org structures.
// Real authority. Real time.

Ready to make your operating model infrastructure?

Join our design partner programme and shape how enterprise organisations run in the AI era.

Become a Design Partner