Manchester Worked from Barcelona UK GDPR · ICO One hour behind us

Manchester.
UK-hosted by default.

Manchester sends us a different sort of problem than London does. It is rarely a regulator asking how a decision was made; it is a catalogue of two hundred thousand products that nobody can search properly, a support queue that grows faster than headcount, or thirty years of broadcast material sitting on a shelf that is technically digitised and practically unreachable.

Last reviewed 5 September 2026 · Engineering guidance, not legal advice.

WHAT IS ACTUALLY DIFFERENT HERE

UK GDPR, and the ICO reads it

A Manchester company sits under the UK GDPR and the Data Protection Act 2018, supervised by the Information Commissioner's Office. The rules read very like the EU ones we work under daily, but they are now maintained separately and can drift. Where your customer base straddles both, the architecture has to satisfy the stricter reading rather than the more convenient one.

Adequacy is renewed, not granted

Personal data moving between the UK and the EEA currently relies on adequacy decisions that are reviewed periodically. We treat that as a live variable: where a system can be built so it does not depend on the decision surviving, we build it that way. It costs almost nothing at design time and a great deal in a hurry.

UK compute is in the south, and that is fine

There is no Manchester cloud region. The major providers run their UK regions in and around London, so keeping data in the UK is straightforward and keeping it in Manchester is not a thing anyone offers. For residency purposes that distinction does not exist; for latency it is a handful of milliseconds, which no document pipeline will ever notice.

Rights, records and clinical governance

Retail brings consumer and advertising rules to any system that writes copy or sets a price. Broadcast and publishing bring rights: an archive is only useful if you know what you are licensed to reuse, and that metadata is usually the hard part. Health technology brings clinical governance and, once software starts influencing care, the MHRA. None of these blocks the work; each decides where a person has to stay in the loop.

THE WORK THAT COMES OUT OF THIS CITY

Manchester and the surrounding towns hold one of the densest concentrations of online retail in Europe, a national broadcast cluster at Salford, a growing health technology corridor around the universities and the hospitals, and the professional services firms that serve all three. The AI problems that arrive from them are mostly volume problems: too many products, too many tickets, too many hours of footage.

Retail and marketplace businesses come to us because their search is losing them money. Someone types a phrase a merchandiser never anticipated, gets nine irrelevant results, and leaves. That is a retrieval and ranking problem long before it is a language-model problem, and we treat it as one — see merchandising and search. The same companies usually arrive with a second, quieter problem: a contact centre whose cost scales exactly with order volume, which is what customer service automation exists to break.

The media businesses around MediaCity have the opposite shape. Their archive is enormous, valuable, and effectively write-only, because finding the ninety seconds you remember means asking someone who was there. Transcription plus retrieval over the archive turns that into a search box. It is the least glamorous project in the building and reliably the one that pays for itself first.

Health technology companies here tend to be small teams with clinical advisers and a very clear sense of what they are not allowed to automate. That is a good starting position. The work is usually documentation and triage support with a clinician holding the decision, which is the pattern described under healthcare and life sciences.

HOW A MANCHESTER PROJECT ACTUALLY RUNS

Discovery happens in your building

Barcelona to Manchester is a direct flight and a short one. The first workshop is worth doing beside the people who actually run the catalogue or the archive, and we do not put that trip on the invoice for a normal engagement.

Then remote, with us an hour ahead

Barcelona runs one hour ahead of Manchester, so our afternoon is your lunchtime and a question raised at nine gets an answer before the standup the next morning. The whole working day overlaps.

Back over for handover

The session where your engineers take the system over is a working day, not a presentation. That one we do in person as well. Afterwards the code is in your repositories and it runs in your accounts.

QUESTIONS

QUESTIONS — 5

No. We have one office and it is in Barcelona. We fly over for discovery, for handover and for anything where standing next to the problem changes what we learn. A studio our size claiming a Manchester address would be describing a registered-office service.

Yes, though the honest version is that it stays in a UK region rather than in Manchester — the providers run their UK capacity in and around London. Within that, storage, inference and logs can all be pinned to the UK, and open-weight models can run inside your own perimeter. Where a hosted model is involved we write down its retention and training terms at the time of writing, and you should check them against the provider's current terms before you sign.

The ICO. It has published guidance on AI and data protection for several years, and its concerns are consistent and unsurprising: a lawful basis for the processing, transparency about what is automated, accuracy, and a route for a person to contest a decision that affects them. We design so those are answerable. Your DPO owns whether the answers are sufficient.

It is the project. Roughly the first third of a retail engagement is usually spent on the data behind the search rather than on the model in front of it, because ranking cannot recover attributes that were never captured. We would rather say that in week one than discover it in week six.

Neither. Our ranges are published on the pricing page, in euros and exclusive of VAT, and they are the same wherever you are. We do not price by postcode.

Keep reading

Industry

AI for E-commerce & Retail

→

Architecture

RAG & Retrieval Systems

30-minute technical call · no deck