Frihet — AI Product Architecture & MCP Integration

Role: Founder and Product Architect. Project type: founder-led SaaS product.

Frihet brings invoicing, expenses, customer records and business operations into one product. My work connects product direction, interface design and implementation with the integration layer that lets AI assistants use business tools.

Small teams often move the same information between documents, spreadsheets and separate applications. Adding an AI assistant is useful only when its actions connect to the underlying workflow and can be checked.

I organised the product around connected business records and clear user journeys, with explicit review points for imports and proposed actions. The public demo follows a single business scenario from an accepted quote through invoicing, payment, reconciliation and accounting, using clearly labelled fictional data.

The open-source Frihet MCP server provides a concrete integration reference: tool schemas, structured outputs, local and remote access, client configuration and a demo mode that simulates writes without touching production records.

Product and information architecture across connected business workflows.

AI-assisted implementation with defined review and verification steps.

MCP and API integration beyond a standalone chat interface.

A coherent relationship between product positioning, interface and operating model.

This case presents the product and publicly inspectable implementation. Demo records are fictional; no customer data is shown.