I build AI-native products with founders who move fast.

A small services company keeps projects in one tool, clients in another, invoices in a third and cash in spreadsheets. Milano brings agile project management, CRM, Spanish invoicing, treasury and financial planning into a single app with a single data model, and exposes it all through an API and MCP so AI agents can do the mechanical work. It started as Collybrix's internal tool and is now becoming a multi-tenant product.

Project documentation goes stale the moment it is written, and the real knowledge lives scattered across the repo, Drive, meetings and the management tool. Parrot ingests all of those sources, generates and regenerates technical and product documentation from code and specs, and answers questions in natural language with mandatory citations. When there is no evidence, it says so instead of making something up.

Running a whole year's budget in a spreadsheet works until a formula breaks, you can't open it on your phone and every January means copying and cleaning the template. Fintio turns it into an app: set up categories and projections once, log each month's transactions, and it works out carry-overs, the 50/30/20 rule, subscriptions, savings and net worth without touching a formula. I built it solo, from product to deployment; it's a public MVP in production since December 2025.

Every day, Río Negro's road authority publishes the state of the provincial routes as a table in a PDF. In winter, anyone driving to Bariloche, El Bolsón or across the Línea Sur needs to know whether a road is passable, whether chains or 4×4 are required, and not to read a document. This app reads that report with AI, turns it into filterable data, draws it on a map and builds a trip planner that analyses road conditions section by section. In Spanish and English, for tourists.