AIReleased · 2025
MaraponeAI
Private, owned intelligence — not another AI subscription.
MaraponeAI is the engine we build the rest of Marapone on: a way to give a company real document intelligence and risk analysis without handing its data — or its budget — to a third-party cloud.
The premise is simple and unfashionable. You buy it once. It runs on your infrastructure. You get the source. Nothing you feed it leaves the building, and there is no monthly meter running against your usage.
- 100%
- On-prem — no cloud dependency
- 0
- Per-seat subscriptions
- 2
- Industry suites shipped on it
01The problem
Every AI vendor selling into construction and logistics wants the same thing: your documents in their cloud, and a seat-based bill that grows forever. For firms whose entire edge is proprietary — bids, drawings, rates, contracts — that is a non-starter.
We wanted to prove the opposite could work: a serious, domain-tuned model that lives entirely inside the customer's walls and is theirs to keep.
02What we built
A private inference stack that runs a quantised llama.cpp engine on commodity hardware, wrapped in a domain layer trained on real operational data — blueprints, RFIs, bills of lading, tender packages.
On top of it sits GasperAI, the assistant that ships inside every Marapone product. Same engine, same privacy guarantee, whether it is reading an Ontario Building Code drawing or auditing a freight invoice.
03The outcome
Two full industry suites — construction and logistics — now run on the same owned engine, in production, on customer hardware. No black boxes we can't open, no dependency we can't replace.
Built with
llama.cppPythonPyQt6Custom fine-tuningOn-prem
More from the lab
Start a project
Have something worth building?
A game, an app, an AI model, or a problem that doesn't have a name yet — tell us about it and we'll build the prototype.