Describe the policy. The agent configures the tenant.
Fingrid exposes its Configuration Studio to AI agents over MCP — the Model Context Protocol. Implementation consultants and client teams describe products, workflows and rules in natural language; the agent translates them into governed configuration, with every change passing the same maker-checker gates a human change would.
Natural language in, governed configuration out
Fingrid MCP server
A first-class MCP interface to the platform's configuration engines — workflow, rule, task and parameter.
Policy-to-configuration
Hand the agent a credit policy document; get back proposed rules, grids and deviation matrices to review.
Workflow authoring
Describe a journey — 'two-wheeler with dealer sourcing and co-lending split' — and the agent drafts the stage flow.
Conversational changes
'Raise the FOIR cap for salaried above ₹1L to 60%' becomes a parameter change request, effective-dated.
Implementation copilot
Consultants configure tenants in a fraction of the clicks — the agent handles the mechanics, they handle the judgement.
Any MCP client
Works with Claude and other MCP-capable agents your team already uses.
Governance is non-negotiable
Agentic configuration changes are still configuration changes:
- Every agent-proposed change enters as a draft — maker-checker approval before anything goes live
- Effective dating, versioning and rollback apply identically to agent and human changes
- The audit trail records that an agent proposed it, from what instruction, and who approved it
- Agents operate within role-scoped permissions — an agent can't touch what its operator couldn't
Watch a policy document become a tenant.
Bring a real credit policy to the demo. We'll configure live.