Make selected workflow capabilities available to compatible AI clients through MCP. Connect AI-assisted experiences to useful actions while maintaining clear boundaries around access.
The Model Context Protocol is an open standard that lets AI clients discover and call tools. Oryntiq turns your workflows into those tools.
One connection, many clients
Expose workflows once and use them from any compatible AI client.
Actions, not just answers
Let AI assistants trigger real, tested workflows instead of guessing.
Clear boundaries
You decide which workflows are visible, who can call them and what needs approval.
When a client connects, it asks the server which tools exist. Each exposed workflow is described with a name, a purpose and typed inputs.
- Clear names and descriptions for the AI
- Typed inputs validated before a run
- Only exposed workflows are listed
// What a client sees after connecting{"tools": [{"name": "search_orders","description": "Find orders by customer or id","inputSchema": { "query": "string" }},{"name": "issue_refund","description": "Refund an order (approval > $50)","inputSchema": { "order": "number", "amount": "number" }}]}
Nothing is shared by default. Pick individual workflows, describe them for the AI and choose which ones need a human in the loop.
- Opt-in, workflow by workflow
- Approval rules per tool
- Turn a tool off without breaking the client
Each client gets its own scoped credentials. Revoke access at any time and see every call in the execution history.
- Per-client tokens with scopes
- Rate limits and allowed tools
- Full audit trail of tool calls
// Scoped access for one AI client{"server": "oryntiq-production","auth": { "type": "bearer", "scopes": ["orders:read", "refunds:write"] },"tools": ["search_orders", "issue_refund"],"approval": { "issue_refund": { "above": 50 } },"rateLimit": "60/min"}
Follow a single request from an AI client through an approved workflow and back.
From an existing workflow to a tool your AI client can use.
Pick workflows
Choose the workflows you want to expose and give each a clear description.
Create credentials
Generate a scoped token for each AI client.
Connect the client
Add the server URL and token to your compatible AI client.
Test a call
Ask the assistant to use a tool and inspect the run in execution history.
What teams usually ask before connecting AI clients.
Any client that supports the Model Context Protocol. Add the server URL and a token, and the client can discover the tools you exposed.
Bring your tools together, automate the steps that slow you down, and put AI to work where it makes a difference.