Bilingual AI Performance Assurance

AI that works equally well in both languages. Verified.

Most AI is never actually tested for parity between English and French, and the gaps are specific: lower accuracy in French, drifting terminology, regulated wording that loses its meaning, bias in one language but not the other. We test for all of it, and give you documentation that stands up to an auditor or a regulator.

Start with a $95 scan → See the tiers

Which one do you need?

Two different jobs, two different products. Pick the one that matches what you have in front of you.

A single document

You have one document to check

You hold an English document and its French version, and you need to know they match, in accuracy, terminology, and legal meaning. Self-serve and fast, with the option to add a qualified human who confirms, corrects, and certifies.

From $95, self-serve
See document tiers →
A whole AI system

You have an AI system to evaluate

You run an AI system producing bilingual output at scale, and you need it evaluated across many prompts. That is the only way to measure bias parity and system-wide behaviour. A scoped engagement, quoted after a short call. A document check is not a substitute.

From $4,500, scoped engagement
See system evaluations →

Built for organizations accountable for both official languages:

Crown corporations & health authorities

Under the Official Languages Act or a bilingual service mandate, verifying AI parity before and during deployment.

Federal contractors & suppliers

Bidders who must show bilingual AI compliance in procurement, with documentation that stands up to review.

AI vendors

Product teams seeking independent pre-market validation of their French performance before pitching public-sector buyers.

For a single document

Check one English and French document

The Automated Parity Scan catches what a machine can check, instantly and self-serve. Each tier above it adds a qualified human. Pick one tier: prices are all-in, and a higher tier already includes everything below it. Pricing in CAD, covering up to and including 1,500 words, then a small charge for each 100 words beyond, so you see your exact price before you commit. Work begins only after payment in full.

Files we accept
Plain text (.txt, .md), Word (.docx), and text-based PDF (.pdf), up to 5 MB and 1,500 words for a standard document. Longer documents are quoted. Text must be selectable, so scanned images need OCR first. Everything is handled as plain text in a controlled environment. We do not open or run uploaded files, and the instant scan reads your text in your own browser without uploading it.

Automated Parity Scan

$95 up to 1,500 words
+ $5 per 100 words beyond
A fast, automated self-check before you decide what you need. No human, no waiting.
What you get
  • Instant automated pre-screen
  • Numbers, dates, amounts, terminology, untranslated text, formatting, obligation-language checks
  • Every issue with what, why, the fix, and a severity
  • Document parity metrics
Turnaround: instant
Certificate: none
Human review: none
See a sample, then buy
Most popular

Reviewed Parity Check

$450 up to 1,500 words
+ $25 per 100 words beyond
You need a qualified human to confirm the issues, fix the French, and vouch for it.
What you get
  • A qualified bilingual reviewer confirms each issue
  • The French is corrected for you
  • A signed review letter you can share
Turnaround: a few business days
Certificate: signed letter
Human review: yes
Order a review

Certified Document

$1,200 up to 1,500 words
+ $60 per 100 words beyond
A regulated or Official Languages Act file that must stand up to an auditor or regulator.
What you get
  • Domain-expert review of regulated wording
  • A Regulated Assurance Seal
  • A public verification-registry entry anyone can confirm
Turnaround: longer
Certificate: Regulated Assurance Seal
Includes the $650 Seal in the price
Order a certification

Already bought a Reviewed or Certified check? Submit your documents →

For an entire AI system

Evaluate a whole AI system

These engagements test the AI system itself across many prompts. That is the only way to measure bias parity and system-wide behaviour, and a document check is not a substitute. Every engagement is scoped and quoted after a short discovery call. Pricing in CAD.

Service Scope Price
Baseline Evaluation Single AI system, up to 5 use cases, four dimensions $4,500
Baseline + Certification Statement Baseline plus a formal, dated Statement $6,500
Annual Monitoring Retainer Quarterly re-evaluation and annual renewal $8,000 / yr
Vendor Pre-Market Validation Full model evaluation, 3 language domains $12,000
Book a discovery call →

What we evaluate

Four dimensions, scored by qualified bilingual evaluators and combined into a single Bilingual Performance Score. The document tiers apply the first three to your file; the system engagements test all four across the live system.

Dimension 1

Linguistic Output Quality

Accuracy, fluency, and completeness in each language, assessed independently and then compared.

Dimension 2

Terminology Consistency

Equivalent concepts and terms across both languages, checked against a domain glossary, with no quiet meaning drift.

Dimension 3

Regulatory Language Equivalence

Whether regulated wording carries the same legal and policy meaning in both languages, confirmed by a domain expert.

Dimension 4

Bias Parity Across Languages

Whether the system treats sensitive attributes differently depending on the language, tested with paired adversarial prompts. Needs the live system.

Verify it before you deploy it

Whether it is one document or a whole system, you walk away with documentation you can put in front of a regulator, an auditor, or a procurement office.

Start with a $95 scan →

The Bilingual AI Performance Statement attests to performance parity between English and French within a defined scope and period. It is a governance document, not a legal opinion, and it does not certify compliance with any specific legislation.