We built the AI infrastructure the UK couldn't get anywhere else.
A managed inference platform for regulated UK industries. Open-weight models on sovereign British infrastructure, with zero data retention guaranteed by architecture, not promises.
The story
Sovereignty became a deal-killer. So we did something about it.
Pendra didn't start as a pitch. It started as the same problem, hit over and over, until we stopped waiting for someone else to solve it.
01 · Where it started
We met building the platform behind an AI company.
For years the two of us led engineering, security and R&D at AMPLYFI. Liam ran product and applied ML; Tom owned the infrastructure and security beneath it. Together we shipped systems that read the open web and made sense of unstructured data at the scale of billions of documents.
02 · The wall
The same wall appeared, again and again.
The organisations with the most to gain from AI were also the most wary of it: healthcare providers, law firms, banks, government departments. Every conversation ran into the same hard questions. Where does our data go? Is it used to train someone else's model? How long is it kept? With every credible provider, the honest answer meant sensitive data leaving for a US cloud. So adoption stalled, not on the technology but at the procurement gate.
03 · The shift
Then the ground moved under all of it.
Microsoft told the French Senate, under oath, that it couldn't guarantee European data stays beyond the reach of US law. At the same time, open-weight models caught up, matching the closed ones on exactly the document-heavy work regulated industries run. Staying sovereign no longer meant settling for weaker AI.
04 · Pendra
So we built the thing we kept wishing existed.
Open-weight models, on UK soil, under UK jurisdiction, with zero data retention enforced by the architecture itself. A drop-in OpenAI-compatible API that lets any UK organisation adopt AI without surrendering control of its data.
The infrastructure we kept wishing existed. So we made it.
— Liam & Tom, co-founders
What we believe
Three principles we won't trade away.
Data residency
Sovereignty by architecture
Zero data retention isn't a policy line you have to trust. It's how the platform is built. With Pendra's managed inference, your data is processed under UK jurisdiction and never leaves British soil.
Model quality
Open weights, no compromise
Open-weight models are approaching the performance of frontier proprietary systems. On the document-heavy work regulated industries run, they're already comparable. Sovereignty no longer means weaker AI.
Built-in controls
Compliance as foundation
Encryption, audit logging, data residency: the controls regulated work depends on are built into the platform from day one. Compliance becomes the thing that gets your project through review, instead of the thing that holds it up.
The founders
Two engineers who hit this wall first-hand.
Liam brings nearly a decade in applied AI and ML to Pendra. He was previously CTO at AMPLYFI, where he led product, engineering and R&D, building technology that reads the open web and makes sense of unstructured data at scale.
His approach is grounded in both academic research and commercial delivery, spanning prior roles as a research assistant, software engineer and UX designer. He holds a PhD in Human-Computer Interaction from Swansea University, where he collaborated with IBM and the BBC. He also runs Creu Labs, a Wales-based venture studio.
Tom leads the architecture of Pendra's sovereign inference infrastructure. An infrastructure and security specialist, he has a track record of building resilient, hyper-scale systems: managing Fortune 500 security environments, delivering critical UK public-sector infrastructure and architecting data pipelines handling billions of documents.
He was previously Head of Infrastructure & Security at AMPLYFI, where he built and scaled the platform underpinning the company's AI and data products. As an engineering leader, he builds security-first cultures that bridge rapid innovation with enterprise-grade compliance.
Where we're from
Built in the UK. Hosted on British soil.
Pendra AI Ltd is a UK-incorporated company. UK jurisdiction isn't a feature we bolted on. It's the whole point. Your data is processed here, governed by UK law, and never leaves the country.
Get in touch
Building AI products for regulated industries?
If you're building an AI product for healthcare, legal, public sector or finance, we're the inference layer behind it. Sovereign, UK-hosted and compliance-ready — so you can ship without owning the regulatory surface yourself. Let's talk.