Auvergne-Rhone-Alpes (ARA), France
AI development for Lyon teams, handed over in writing
We build AI products from Vancouver, British Columbia, and we work with Lyon teams without a single shared working hour. No office there, no staff there. What replaces the shared day is a written handoff, decisions logged as they are made, and review that happens on your morning rather than in a call.
- base
- vancouver, bc, canada (UTC-08:00)
- market
- Lyon, France
- your clock
- UTC+01:00 (Europe/Paris)
- overlap
- none on the same date, written handoff instead
- presence
- remote only, no office in Lyon
No shared working day, so the loop is written
Lyon is 9 hours ahead of Vancouver. When we start at 09:00 it is 18:00 on the same day for you, and when we stop at 17:00 it is 02:00 on the next day. On the same calendar date there is no hour where both sides are at work.
There is a real cost and it is worth naming: an answer to a blocking question can take a working day to reach you. We handle it by sending the proposed answer with the question, so silence means agreement rather than a lost day.
- The Vancouver day ends with a written summary: shipped, broken, blocked, and the one thing we need you to decide.
- That summary is waiting at 02:00 your time, alongside the diffs it refers to.
- The decision log lives in the repository rather than in a chat thread, so it is still readable a year later.
- Anything that would block us gets asked as a proposal you can approve in a sentence, not as an open question.
One live call a week, and it is scheduled honestly: 08:00 in Vancouver is 17:00 in Lyon. That hour is outside our working day and inside yours. We take the awkward end of it, because it is our job to be reachable.
What Lyon buyers usually bring us
The record for Lyon lists two industries. Here is the part of each one that turns into an engagement, and the service line it lands under.
pharmaceuticals
Regulatory documents and study reports have strict structure and strict traceability, so parsing that structure is the bulk of the schedule.
Answers cite the passage they came from. A retrieval system without span-level citation gets used twice and then quietly abandoned.
Closest service line: Knowledge bases and retrieval
biotech
Protocols, study reports and regulatory correspondence answer the same questions repeatedly. Structure-aware ingestion of the report format is most of the work.
Most retrieval failures are ingestion failures, so the early weeks go into parsing and chunking rather than into rewording a prompt.
Closest service line: Knowledge bases and retrieval
Contracts, currency and where the data sits
Invoices are in euros. GDPR makes storage location a design question rather than a footnote, so the index, the logs and the model calls get pinned to European regions where your provider offers them, and that is settled in week one.
Beyond that, the shape of an engagement does not change with distance: a scoping call, a thin end-to-end slice, then weekly increments against acceptance criteria fixed in advance. If we are not the right shop for Lyon work, you get told that in the first reply rather than after a proposal.
The five lines we sell
Same five lines everywhere, Lyon included. Most engagements are one of them, sometimes two.
- AI products, built end to end
- Knowledge bases and retrieval
- Codebase and team optimization
- Training and enablement
- Forward-deployed engineering
The languages, models and infrastructure underneath all five are listed on the Magic Ship stack page, with what each one is good for and where it is the wrong choice.
Other markets in Europe
More of Europe, all of it worked remotely from Vancouver.
Start a project
Send the problem rather than a job spec. You get an answer on scope, timeline and whether we are the right shop, written by the person who would do the work.