A personal AI project

What we build,
with source and measurement.

Neuralex is a personal project. This site shows what runs here: AI systems with source attribution on every answer, built in our own lab — you can check it yourself.

Read on

A language model sometimes makes things up. So we make it checkable.

In our systems every answer carries a clickable source, and the hallucination rate is measured against a fixed set of questions with known answers. What we claim can be rechecked — and what we don't (yet) know is stated too.

How we work

Reliable AI is not a promise but a design

Proven, not promised

Everything shown here is already running in our own lab. No slides — working products you can inspect yourself.

A source with every answer

Every answer from our systems is clickably traceable to its source. The hallucination rate is measured against a fixed golden set — measured, not hoped.

Data stays where it belongs

Our systems run on our own hardware in the Netherlands, behind a Cloudflare tunnel. We tell you exactly which part runs where.

The proof

Look inside the lab

We show what we build — from first idea to running product, including the technology behind it. You can see for yourself.

Keep looking

See the projects

Every project page describes what it is, how it was built and what we measured — with a link to the live product.