Insights
Notes for the people who sign off.
What we have learned building AI for institutions that are held to account — procurement, sovereignty, compliance and the engineering underneath.
How to buy AI as a Canadian public body
Most AI procurements fail before the RFP is written. A working sequence for Canadian public bodies: the decision, the assessment, the vehicle, then the vendor.
7 min read
RAG vs fine-tuning: which one for policy documents?
Retrieval answers from current text and cites it; fine-tuning changes how a model writes. For policy documents you usually want the first, sometimes both.
6 min read
What “sovereign AI” actually means for data residency
Residency is where data is stored. Sovereignty is who can be compelled to hand it over. The difference decides which architecture a public body can accept.
6 min read
Evaluating an AI vendor: the audit checklist
Twenty questions that separate an AI vendor who can be audited from one who can only present. Use them before shortlisting, not after a contract is awarded.
8 min read
What PIPEDA means for your AI vendor
Outsourcing processing does not outsource accountability. What Canada’s private-sector privacy law requires when personal information goes to an AI vendor.
6 min read
Have a procurement coming up?
We brief evaluation teams before an RFP is written, not only after.
Request a Briefing