We are based in Barcelona and work with Roman organisations remotely, on the same clock. Rome is not Milan with monuments: the buyer here is usually spending public money or answering to someone who is, which means the question is rarely whether a system works. It is whether you can prove afterwards what it did and why.
The organising fact about Rome is scale of correspondence. Ministries, agencies, national utilities and telecoms operators all run enormous flows of documents between themselves and the public, and most of the useful AI work in the city is somewhere in that flow.
Public administration brings citizen-facing document processing: applications, claims, certificates, appeals. The failure mode here is specific and worth naming — a confidently wrong answer to a citizen is not a bad customer experience, it is a decision someone may act on and later contest. So these systems get built to assist a named official rather than to replace one, with the source document quotable in the response. That is the substance of our public sector work.
Energy and utilities arrive with regulatory filings, grid and asset records, and a long tail of technical documentation that engineers currently search by memory. Retrieval over that is unglamorous and it is usually the highest-value thing available.
Defence and aerospace organisations arrive with the strictest deployment constraints of anyone we speak to, and often the answer is an open-weight model running on their own hardware with no external call at all. We describe what that costs and what it gives up under on-premise AI, rather than pretending it is free.
Telecoms bring volume: support contacts in the millions, where even a small deflection rate is material and a small error rate is a regulatory problem. The engineering is in the routing and the refusal behaviour, not the fluency.
Across all four, the deliverable we are most often asked for after the system itself is the record of how it behaves over time, which is why observability tends to be in scope from the start here rather than added in year two.
No. We have one office and it is in Barcelona. We travel to Rome for discovery, for handover and for anything where being in the room changes the outcome. We have no Italian entity, no local partner firm and no representative here, and we would rather say that at the start than have it emerge during a procurement check.
Be careful with how you use us here. We are a small engineering studio without an Italian entity, so we are not a realistic prime bidder for a public tender and we will not claim otherwise. Where we are useful is as the technical supplier or subcontractor to an organisation that is bidding, or working directly with a body under a threshold or a framework it already holds. We can also write the technical sections a bid needs — architecture, data flows, evaluation approach, oversight design — which is often the part that is hardest to get right.
Yes. The Italian cloud regions are in the north rather than in Rome, which changes latency rather than jurisdiction, and open-weight models can be self-hosted on infrastructure you already control, including with no external network path. Where a hosted model is used we document its retention and training terms at the time of writing, and you should confirm them against the provider's current terms.
That depends on what the system decides and about whom, and it is worth establishing in the first week rather than the last. Broadly: the more directly an output affects a person's access to a service or a right, the more the obligations shift towards documentation, human oversight, logging and transparency. We build so those are available; your legal function owns the classification and the declaration. We are engineers, not counsel, and a page like this is not the place to be told which category you are in.
The same as anywhere else — the ranges are on the pricing page, in euros and exclusive of VAT. We do not price by postcode. Where a project carries heavy documentation and audit obligations the estimate goes up, because that is genuine work, and we would rather show it as a line than hide it in a day rate.
Thirty minutes on the problem, and an honest read on whether we are the right team for it.
30-minute technical call · no deck
What we built, what stage it is at, and a link to the live thing wherever there is one.
yourOS is an operating system with an AI at its heart. You talk to it like a person, out loud or typed. It makes your calls, plans your days and works on your projects with a crew of agents, on your own private server. Our own product, built from the first commit; the waitlist is open.
Website, backend and app assets for Sonora, a free AI sound-therapy app on iOS and Android published by Cognitive AI.
Brand, website, prototype, apps on five platforms, the AI backend and the algorithm-intelligence pipeline for Podswap, a social growth engine that runs without a human in the loop.
In closed beta: Roxy, a private AI assistant with permanent memory running on infrastructure the user owns. Brand, site, product, backend and conversational design.
Live at coherence.com: brand identity generated with AI agents we directed, the motion system and website for the group behind Sonora, AboutAI, WakeUp and MudShadows. Agent work remains in build.
Live at domainbooth.com for DomainBooth, a premium domain brokerage: a new animated mark, every page redesigned and rebuilt with designed mobile layouts, and a custom CMS for its sales record, featured names, press and enquiries.
Apps, platforms and campaigns for these teams, long before the AI work. Every one is written up on the work page — what we did, and where our part ended.
