We are building the expense data layer that your AI agents will understand.
ONexpense already extracts expense data with frontier language models. We are building the API and MCP server so your agents can read, write, and natively understand expense data.
What you can do today
ONexpense uses frontier language models — not legacy OCR — to automatically extract data from expense receipts: amount, VAT, category, date, currency, merchant, payment method, and merchant location.
The Sur-Mesure plan includes custom third-party API connectors, managed by our team. If your company needs to integrate ONexpense with an ERP, HRIS, or existing accounting workflow, contact us for a personalised demo — we build the connector for your system.
Request a Sur-Mesure demoThe MCP server and public API
Our MCP server is in production, in private beta. Join the waitlist to request access. The self-serve public developer API is in development — public availability planned for 2026.
We are designing the ONexpense API so your AI agents can read and write expense data without a custom adapter layer. Here are the design principles we are building toward:
- → LLM-friendly schemas
Semantic field names (
merchant_name,vat_amount,receipt_category) that agents can consume without manual mapping. - → OpenAPI 3.1 specification
A machine-readable spec for direct agent introspection and automatic code generation.
- → MCP server (Model Context Protocol)
Expense data as a native MCP resource — your AI assistants will be able to read receipts, create reports, and trigger submissions from their workflows.
- → Agent-readable error messages
Structured errors with natural-language descriptions — your agents understand what happened without custom parsing.
Join the waitlist
Beta partners get immediate access to the MCP server (private beta) and directly shape API decisions — field naming, MCP resource granularity, use-case priority. If you are building an accounting agent, an automated ERP workflow, or an expense orchestrator, this is for you.