Model Context Protocol

Chat with your process knowledge

Crismo ships an MCP server so AI agents like Claude, Cursor, and Cline can answer precise questions about how your organization actually works, read straight from your process maps rather than guessed from documents.

Claude · Crismo MCP
If a support ticket is escalated and the customer is enterprise-tier, who handles it?
Enterprise escalations route to the Account Manager for that customer. If no AM is assigned, it goes to the Enterprise Support Lead. This comes from the Support Escalation diagram, an exclusive gateway on customer.tier == "enterprise".View in process map

Answers are read directly from your structured process maps, not guessed from prose.

Structured, not fuzzy

A graph of how work happens, not a wall of text

When you map a workflow in Crismo (or the AI extracts one from a meeting transcript), you capture explicit relationships: who does each step, what triggers it, where it branches, and how it connects to other processes. The MCP server hands that structure to your AI agent, so answers are precise instead of hedged.

  • Ask in plain language, in the tool you already work in
  • Traverse real decision logic: conditions, branches, loops
  • Follow handoffs and dependencies across linked diagrams

Document search vs Crismo MCP

Document search
“Enterprise customers receive priority support… escalated tickets go to senior agents…” (you piece it together)
Crismo MCP
“Enterprise escalations route to the Account Manager. If none is assigned, the Enterprise Support Lead.” (traversed the actual logic)

Trust by design

Every answer traces back to a visual map

This is what a wiki page can never do. When the agent says 'enterprise escalations go to the Account Manager,' one click opens the actual process flow: the exact step, what comes before and after, who else is involved. Answers aren't AI hallucinations; they're direct reads from an authoritative model your auditors and teams can see.

  • Governance: auditors see the real logic, not a summary
  • Maintenance: update the map, the knowledge updates everywhere
  • Tenant isolation, permission-aware, tokens never logged

The Operational Truth Layer

Your process maps become the single source of truth for how work actually happens: expressive enough to capture real complexity, visual enough to understand at a glance, and structured enough to be queried by AI.

What agents can do

Turn your process library into an answer engine

The Crismo MCP server gives your AI assistant a set of tools for reading and reasoning over your processes.

Query process knowledge

Ask who does what, in what order, under what conditions, answered by traversing the actual decision graph.

Compare & trace flows

Compare how two teams or regions handle the same process, and trace handoffs across linked diagrams.

Find gaps & owners

Surface steps with no owner, dead-end branches, or missing documentation across your whole library.

Impact analysis

Ask what changes if you move a threshold or a step, before you touch the process.

Getting started

Works in the tools you already use

The Crismo MCP server is provisioned for Enterprise workspaces. Once API access is enabled, we share the server and a ready-to-paste config for Claude Desktop, Cursor, Cline, or Zed, pointed at your workspace. It works with any MCP-compatible client.

MCP, explained

Give your AI agents a source of truth

See how the Crismo MCP server turns your processes into precise, traceable answers your whole team, and your agents, can trust.