Small models, measured honestly, owned by the people who use them.

Why we exist

Every AI system is full of small decisions. Is this message safe? Which team should handle it? Can this agent issue that refund? Is this answer supported by its source? Today most of them are made by sending data to a large hosted model and parsing whatever text comes back.

We think a decision deserves better: a model built only to decide, small enough to run where the data lives, honest about its confidence, and measured before anyone relies on it. That is what Myles is.

Founder

Charlie HudsonFounder

Charlie founded Launchables in 2026 to build decision models that organisations can own, measure and trust: small enough to run on their own hardware, honest about their confidence, and built in the UK.

Get in touch 

What we believe

1. A decision deserves its own model.Built only to decide, a model can be small, fast and excellent at it.
2. Every claim is measured.Frozen test sets, published methods, and our own results labelled as ours.
3. Intelligence runs where the data lives.On the laptop, phone or server you already own, including disconnected networks.
4. You should know where a model came from.A lineage record with every model: base, data, licences and who made them.
5. People decide the rest.What Myles cannot decide within its error rate goes to a person, with the probabilities attached.
6. Build it here.Trained and evaluated in the UK, on UK-owned hardware and UK public text.

What we build

Where we are

• Based in the UK. Formed in 2026. Our models are trained and evaluated on UK-owned hardware.
• Open to pilot partners. Public services, regulated firms and AI product teams who want to measure Myles on their own decisions.
• Shipping in the open. What we release and what we learn goes in our notes