DGF Consult
The working prototype

GC Procurement Advisor

The first working application prototype of the Policy-Anchored AI framework — tested against Government of Canada procurement scenarios, and available for structured walkthroughs and pilot discussions.

01Why procurement

The hardest test for the framework.

Procurement is the proving ground because it stress-tests the framework against everything at once: policy thresholds, trade agreements, security and privacy triggers, procurement vehicles, accessibility, official languages, governance steps — and human accountability.

If the framework can produce a defensible, traceable result there, it can generalize to other compliance-heavy domains.

02What it does

The framework, instantiated.

An informal request becomes structured facts, resolved against authority, into a traceable path forward.

01 · Structured facts

The intake, made precise

  • Commodity
  • Estimated / lifecycle value
  • Security level
  • Personal information
  • Duration & users
02 · Authority resolved

What actually governs it

  • Government Contracts Regulations
  • Trade agreements (CFTA · CETA · CPTPP · WTO-GPA)
  • Procurement vehicles
  • Governance triggers
03 · Traceable output

A defensible path

  • Structured requirement
  • Obligations triggered
  • Open questions surfaced
  • Every claim linked to policy
GC Procurement Advisor — the framework instantiated for Government of Canada procurement. The procurement officer remains the decision-maker.
03What it proves

The three disciplines, in practice.

01

Facts, not assumptions

It separates what was provided, inferred, missing, and provisional — and asks before it concludes.

02

Authority, resolved

Trade thresholds, security and privacy triggers, vehicle eligibility, and governance obligations are resolved against a structured policy model.

03

Traceable output

Every conclusion links back to the policy element behind it — reviewable, reproducible, defensible.

04A worked example

From a sentence to a defensible path.

“We need a cloud workflow tool for 200 users to manage internal case review tasks. The system may handle Protected B information including personal information. Estimated cost is $180,000 per year for three years.”

A general-purpose chatbot produces something plausible and untestable. The advisor produces something different in kind: it structures the request, then surfaces the obligations the facts trigger — a privacy assessment for the personal information, security and data-residency considerations for Protected B, trade-agreement thresholds from the lifecycle value, the vehicles the delivery model points to.

Where the facts are incomplete, it says so and asks — rather than baking a silent assumption into a confident answer. Every conclusion links back to the policy element behind it.

05Said plainly

What this is — and what it is not.

  • A working prototype and first implementation of the framework
  • Tested against Government of Canada procurement scenarios
  • Available for structured walkthroughs and pilot discussions
  • Decision support for procurement officers, who remain the decision-makers

It is not a finished commercial product, and is not represented as production-certified for Protected B. It does not replace procurement officers, and the language model is never the authority for a compliance conclusion. The framework is demonstrated in procurement; it is not claimed as proven across every domain.

See it for yourself

Request a walkthrough, or discuss a pilot.

A structured walkthrough against real intake scenarios — or a 90-day pilot proposal for organizations that want to test the pattern on their own work.