About Us

We’ve built the process that lets you use the data you couldn’t share.

Built by a team with experience bringing complex AI platforms into enterprises. Founded in 2026, with offices in London and New York.

Two identical limestone slabs standing face to face

Beginnings

Where this started

Nakato started from enterprise AI projects that kept hitting the same barrier: access to sensitive data.

The models were capable, and the use cases were ready. However, sensitive data had to be sanitised before touching the tool. What was left behind was data being tested that didn’t look anything like the original. The results weren’t useful, and pilots never reached production.

Teams repeatedly faced the same choice:

Expose sensitive data, or lose the context? We thought this was a false choice.

So we decided to build the alternative.

Change the face. Keep the shape.

The team

A team that has built and sold together.

We’ve worked together bringing AI platforms into enterprises.

As a team, we’ve built a solution to the problem we kept experiencing. We wanted to approach this in a different way to what we had seen before. To us, combining optimising data for fidelity and privacy meant navigating each dataset and each task individually, understanding the properties required and which details needed to stay and which to be removed, and building a system that generated the right equivalent for each one.

A combination of industry experience, engineering innovation, and linguistic research forms Nakato.

Both steps of our pipeline, detection and then twinning, rely on a holistic comprehension of the linguistic properties of identifiers. Callum guides this approach. Sam has constructed the engine that puts this into operation. He has built it so that it can be deployed on-premises, tailored to each industry’s domain. Tom’s experience as part of the founding team at Quantexa and Suade Labs, and leading growth at UnlikelyAI scales the business from idea to execution. Saskia is Founder’s Associate and Aqua leads partnerships.

What we’re building

What we’re building and why

Nakato is a privacy layer that transforms sensitive data into semantic twins. These twins have equivalent context but a different identity, designed so the model has no reasonable means of identifying the person behind them. They are sent to the model and the answer is restored on return, so your real data never leaves your environment.

Nakato can be integrated into a company’s architecture so that sensitive data can be used with the AI tools you want. We can also transform data for data providers, de-identifying it so it can be shared under data-protection rules while keeping its context.

Our backgrounds in linguistics, engineering, and enterprise sales mean we know how AI projects actually need to work: deployed on-premises, with every bit of the process auditable. We don’t believe you should have to make any sacrifices: not the performance of your tool, nor the safety of your data.

How it Works · See the benchmarks · Assurance

Backed by

  • Plug and Play
  • MS&AD