AI development company for Québec City insurance, Government of Quebec ministries, and Laval-adjacent research.

We do not operate a Québec City office. Aiinfox runs from our Mohali, India HQ and a Frisco, TX office. For Québec City clients, our Mohali team picks up your morning (6:30pm IST is 9am ET) and our Frisco pod covers your afternoon via a US-Hours rotation. For on-site engagements at the parliamentary precinct, Grande Allée, the iA Financial Group HQ, the Beneva offices, the Université Laval campus, or the CHU de Québec sites (kickoff, milestone reviews, security walk-throughs), we travel to Québec City on a scheduled cadence rather than maintaining a sub-scale local presence.

Materially, and the changes are wired into the engagement architecture from week one rather than retrofitted. Law 25 obligations we map onto every Québec City engagement: (1) explicit privacy-impact-assessment requirement before any personal information is processed; (2) cross-border transfer impact assessment before any inference endpoint outside Quebec is used — the default for ministry engagements is in-Canada through every step of the pipeline, including inference; (3) explicit-consent layer for any use of personal information in automated decision-making, with a right to receive an explanation and to contest the decision; (4) explicit-consent layer for biometric data and CAI biometric registry filing for any biometric collection; (5) breach notification to the Commission d'accès à l'information du Québec (CAI) and to affected individuals. For automated-decision-making systems specifically, every model decision is logged with feature attribution suitable for a human-reviewable explanation — this is a system architecture requirement, not a documentation afterthought.

Honest answer: we deliver customer-facing AI in Quebec French (fr-CA) and Canadian English with a native-fluent francophone reviewer on the output, not a native francophone engineer on the development team. For voice agents, we use Deepgram for fr-CA STT and ElevenLabs Quebec French voices for TTS, validated against native Quebec speakers (not generic fr-CA) before any production release. For chat agents and RAG, the model handles fr-CA and en-CA generation; we run a Quebec French reviewer over a representative sample of customer-facing output during week one of the build, then again before launch, and continuously on a sampled basis post-launch. For document intelligence over bilingual corpora, we use models with strong French coverage and validate extraction accuracy against a held-out Quebec French test set. If your engagement requires a fluent francophone engineer embedded in the build rather than a reviewer on the output, we will be straight on the first call and recommend a Québec City or Montreal-based bilingual delivery partner.

Yes. For Industrielle Alliance, Beneva, Promutuel, and Capitale-Nationale insurance operators running underwriting copilots, claims-triage agents, fraud-flagging systems, or eligibility tools, Law 25 imposes a right-to-explanation obligation: any individual subject to a decision based exclusively or principally on automated processing of their personal information has the right to be informed of the principal factors and parameters that led to the decision, and to contest it. We wire that into the system architecture from week one — every model decision logged with feature attribution suitable for a human-reviewable explanation surface. Humans approve everything that touches a regulated outcome, and the explanation surface is built before the model is exposed to a real claimant or policyholder.

Your call, but for Québec City we default to AWS ca-central-1 (Montreal) because it is the closest in-Canada region with full AWS service coverage including Bedrock for Claude inference. For Government of Quebec ministry engagements and any engagement where the cross-border transfer impact assessment under Law 25 requires no data crossing the border, we self-host Llama 3 on vLLM inside the customer's Canadian VPC for zero third-party inference. Azure Canada Central (Toronto) and AWS ca-west-1 (Calgary) are supported. No Quebec personal information silently crosses the border unless your DPA and Law 25 cross-border transfer impact assessment explicitly permit it.

Yes. PIPEDA applies federally to commercial activity, Law 25 imposes the Quebec-specific obligations described above, and the Act respecting health services and social services (the Quebec health information governance regime) governs the management of health information by Quebec health and social services establishments. For CHU de Québec-Université Laval-adjacent, RAMQ-adjacent, and broader Quebec health-system-adjacent healthtech, BAAs or equivalent DPAs are signed before any PHI is shared; deployments default to AWS ca-central-1 inside customer-controlled Canadian clouds; every model and tool call is audit-logged with input, output, prompt version, and operator identity — exportable for the CAI or the contracting establishment's privacy officer.

Most v1 engagements at Aiinfox land between CAD 35,000 and CAD 160,000 fixed-price for a focused build — an AI agent, a RAG system, a voice pipeline, or a bespoke ML model. Larger multi-quarter engagements with fine-tuning, Law 25-aligned compliance work, automated-decision-making explanation surfaces, or Quebec ministry-grade in-Canada deployments typically reach CAD 200,000 to CAD 400,000. Bilingual Quebec French + Canadian English delivery adds roughly 10 to 20 percent depending on the depth of native-reviewer engagement required. Pilots are CAD 12,000 to CAD 28,000. Québec City clients invoice in CAD via bank transfer.

Yes — takeover audits are routine. Step one is reading the code, the data pipelines, the eval results (if any exist), the prompts, the cost telemetry, and the Law 25 documentation (if any). Step two is shipping the smallest valuable change to prove we understand the system in a way the previous vendor did not. Step three is the longer-term plan. Most Quebec takeovers we have seen did not need a full rewrite; they needed evals, guardrails, observability, Law 25 right-to-explanation surfaces, and a senior bilingual engineer who stayed on the build past the discovery phase. We will be honest on the first call about which is the right move.