Bilbs is based in Québec City, with the office in Sainte-Foy. We install a private AI server in your law firm’s office, trained on your firm’s own files. Your lawyers ask it instead of pasting client matters into ChatGPT. Nothing leaves the building. Ever. The managing partner gets one person to call — me — not an agency switchboard.
I spent six years building AI systems inside Canadian companies. The last two of those years I spent almost entirely redoing work that had been built too fast, by the wrong people, in the wrong country. The pattern was always the same: an outside service, client data quietly leaving the building, and a legal team finding out about it after the fact.
Then Loi 25 came into force, and one industry stood out. A Québec law firm lives under strict professional-secrecy rules. Its clients now send formal questionnaires about how AI touches their files. And its own lawyers are already using ChatGPT — in browser tabs the firm can’t see. The missing piece wasn’t the AI. It was a way to keep the AI inside the firm, on the firm’s own server, trained on the firm’s own files, with every question logged.
So I made that the whole business. We install a private AI server in your law firm’s office. We train it on your firm’s own contracts, memos, and matters. Nothing leaves the building. The person you talk to on the first call is the same person writing the code, the same person sitting with your IT director to install the server, and the same person on call when something breaks two years from now. That person is me.
Previously: senior AI engineer at a Canadian SaaS firm, ML engineer at a regulated-healthtech company, researcher at a Québec AI lab. 40+ production AI deployments, mostly in regulated industries. Based in Québec, QC. French · English.
The server lives in your office. The AI is licensed to your firm. The guidebook lives on your IT’s shared drive. If Bilbs closes tomorrow, nothing your lawyers use today stops working.
Your files never leave the building. They never go to another country. They never go to another company — not to Microsoft, not to OpenAI, not even to us. The answer to every client questionnaire is one page long because the architecture made it one page long.
The lawyer clicks through and reads the source before anything goes out. The same review a senior partner does on a junior’s draft today — just on a draft that’s 90 seconds old instead of four hours old. The AI is a first drafter, never a decider.
One line on the invoice: from $29 per lawyer per month. Goes up only if the firm chooses extra services. Hardware is audited separately — yours if you have a server, our OEM-cost quote if you don’t. The partnership signs the price once.
The person on the first call is the same person writing the code, installing the server, and on call two years from now. No project managers. No team overseas. No juniors getting lost in a document system they’ve never seen.
Standard servers. Standard models. A guidebook your IT director can actually read. Nothing clever enough to surprise anyone at 2 a.m. — because surprises at 2 a.m. are how a firm ends up with the wrong answer in front of a client.
No deck. No pitch. You tell me about the firm; I answer your direct questions about Loi 25, what your associates are already doing with ChatGPT, and what this actually costs. The call is free, and so is the audit that follows it. One line on the invoice: from $29 per lawyer per month — below Microsoft Copilot. Goes up only if the firm asks for extra services. Hardware is audited separately: if you have a GPU-capable server, we use it; if you don’t (no IT yet, no server room yet), we source and install one for you on a transparent OEM-cost quote — your call once you see the spec.