Skip to main content
Everything your company’s money has done is already in ThinkOut — every transaction in and out, every account, every counterparty, sorted into categories and labels that mean something to your business. So is what you plan for it to do next, in your forecasts. Now you can ask about it in plain language. Not just what the numbers are, but what sits behind them: which clients are slowing down, where a cost crept up, whether the month is going the way you planned it, what the balance will be at the end of the year. The kind of answer that makes a decision feel less like a guess. If you want to ask questions in plain language, connect ThinkOut to an AI assistant. You sign in with your ThinkOut account, no keys involved, and from there you just ask. If you want the data inside your own systems, use the REST API. An API key bound to one workspace, and from there your BI tool, your ERP, the script that runs every Monday morning — whatever you already have.

MCP server

Ask about your company’s money in plain language, from Claude, ChatGPT, Claude Code, Codex or Gemini. You sign in with your ThinkOut account — no API keys.

REST API

Pull balances, transactions, forecasts and cash flow reports into your own tools with an API key bound to one workspace.

​
The same numbers, either way

Both routes are read-only. Nothing can be edited, categorised or paid for from here — this is a window onto your data, not a set of controls. And both return figures computed by ThinkOut, using the same rules as the cash flow screen: transfers between your own accounts excluded, split payments counted once, foreign currency converted at the rate on the movement date. So a total you see in a conversation, the same total in your dashboard, and the same total in the app are one number, not three that happen to be close. That consistency is the point. It is what makes the data safe to build on.

​
Where to start

If you are setting this up for yourself, start with the MCP server — it takes a few minutes and needs nothing technical. If you are building something, start with authentication, then the data model, which explains how accounts, movements, categories and labels fit together before you write your first request.