MCP & API

Work in Clive, or from Claude, ChatGPT, or your own code.

Everything you can do in the Clive interface, you can do through the public API or through MCP. You get the rigor of a modeling system you did not have to build, and you still work in the tools your team already uses.

Try Clive free

One model. Three ways in.

MCP

Ask about your numbers from Claude or ChatGPT.

Connect Clive to the AI tool your team already works in. Ask a question there and the answer comes from your live model rather than from a file someone exported last week.

  • Query metrics and list dimensions without leaving the tool you are in
  • Run a workflow, trigger a data refresh, or update a dashboard from the same conversation
  • Skip the step where you paste context into a prompt before you can ask anything

Answered from the live model, inside your own permissions.

Public API

Move data in and out on your own schedule.

The public API covers create, update and delete, the same actions available in the interface. Write the integration you need instead of waiting on a vendor-built one.

  • Push actuals in from a system that has no native connector yet
  • Pull model output into a warehouse, a BI tool, or a script you already run
  • Automate the parts of your monthly refresh that repeat every cycle

RESTful, with developer docs.

Permissions and audit

Connect as yourself, not as a service account.

Users connect to your external AI tools as themselves, with only the permissions they already have in Clive.

  • Keep destructive actions blocked by default
  • Confirm anything that creates or updates before it happens
  • Read the audit log for every action an AI tool took, the same as for a person

The log records every AI action next to the person who authorized it.

For prebuilt connections to your ERP, HRIS and CRM, see Integrations →

Connect Clive to the tools you already work in.

Start free, then make your first API call or connect your AI tool.

Try Clive free