PAICE Learns to Speak Spanish, French, and Portuguese

A compliance officer in Mexico City needs to prove her team can collaborate safely with AI. A claims adjuster in Rio de Janeiro wants to measure his verification habits before his firm rolls out a new underwriting model. A policy analyst in Quebec City is preparing for a cohort assessment, but her strongest professional instincts live in French, not English.

Until today, all three faced the same barrier: the only way to measure People+AI collaboration capability was in English. That meant professionals either took the assessment in a second language, adding cognitive load that had nothing to do with AI collaboration skill, or they didn't take it at all.

Today, the PAICE (People + AI Collaboration Effectiveness) assessment flow is available in Spanish, French, and Portuguese.

What's New

The PAICE assessment now preliminarily three additional languages across their major regional variants:

Spanish covering Latin American Spanish (es_MX)
French covering Canadian French (fr_CA)
Portuguese covering Brazilian Portuguese (pt_BR)

Beta Notice: Multi-lingual support is currently in Beta. We are actively looking for native speakers to try the assessment and share their feedback. If you notice anything that feels off — a phrase that sounds unnatural, a translation that misses the mark, or anything else — we want to hear from you. Try it in your language: Spanish · French · Portuguese.

This is the same five-dimension behavioral assessment that has been available in English since launch and in Urdu since March. Same methodology. Same scoring rigor. Same privacy protections. The difference is that professionals can now complete the entire assessment experience in their native language.

Why This Matters for Organizations

Businesses operating across the Americas and the world share a common challenge: they need consistent, comparable AI capability measurement across linguistically diverse teams. A financial services firm with offices in New York, Montreal, Panama, and São Paulo cannot build a meaningful AI governance program if only their English-speaking employees can be assessed.

Language barriers in assessment create two problems. The obvious one is coverage. If your assessment tool only works in English, you are measuring a subset of your workforce and hoping it represents the whole. The less obvious problem is signal quality. When a professional takes a behavioral assessment in a second language, you are measuring a blend of their AI collaboration skill and their English proficiency. Those are different capabilities, and conflating them produces data that is harder to act on.

Native-language assessment solves both problems. Coverage expands to every professional in the organization. Signal quality improves because the assessment captures authentic behavioral patterns without the noise of language translation happening in the professional's head during every interaction.

With Spanish, French, and Portuguese, PAICE.work now covers the native languages of every country in the Western Hemisphere, and nearly 3 billion of the over 8 billion people across the globe.

PAICE assessment interface shown in Spanish, French, and Portuguese

What Stays the Same

Everything that matters about assessment quality remains identical across languages:

Scoring. The same 0-1000 scale. The same five dimensions: Performance (P), Accountability (A), Integrity (I), Collaboration (C), and Evolution (E). The same dimension weights (A=30%, I=25%, C=20%, E=15%, P=10%). A score of 700 in Portuguese means the same thing as a score of 700 in English.

Methodology. The same behavioral observation approach. PAICE measures what professionals do when collaborating with AI, not what they know about it. The assessment creates conditions where genuine AI limitations become visible and observes how the professional responds. That methodology works the same way regardless of the language the conversation happens in.

Privacy. The same privacy-by-architecture protections. PII is stripped before any language model processes your content. Conversation data is not stored in production. Individual scores are never disclosed to anyone else. These structural guarantees apply in every language.

The free tier. The PAICE assessment remains free in all supported languages. Always.

Please note that this blog post and other parts of the PAICE.work website are still only available in English at this time. What has changed is that the assessment and feedback experience is now localized. We expect to translate all of this website soon, after we incorporate improvements in the assessment itself and graduate from beta mode to fully supported.

How It Works

For Individuals

Select your preferred language when you begin the assessment. The entire experience, from the initial prompt through the AI interaction to your results, will be in that language. No switching back and forth. No English fallbacks midway through.

If you have already taken the assessment in English, you can take it again in another language. Your scores are independently valid and comparable.

For Organizations

Cohort assessments now support mixed-language teams natively. When you enroll participants in a cohort, each individual selects their preferred language at assessment start. The experience is fully localized for each participant, but aggregate reporting works across all languages seamlessly.

This means a single cohort assessment can include English-speaking employees in your New York office, Spanish-speaking employees in your Mexico City office, French-speaking employees in your Montreal Office, and Portuguese-speaking employees in your São Paulo office. Each person gets a native-language experience. Your organization gets a unified view of AI collaboration capability across the entire team.

Individual privacy protections apply regardless of language. Employers see aggregate distributions and dimensional patterns, never individual scores, in any language.

The Road Here

We made a deliberate architectural choice when we began multilingual work earlier this year: start with the hardest language first.

Urdu, which launched for testing in March, is a right-to-left, non-Latin script language. Building for Urdu first forced us to solve the most challenging problems in our detection and scoring pipeline upfront, including Unicode handling, bidirectional text processing, and script-agnostic pattern matching. The architecture that emerged from that work made adding Latin-script languages significantly simpler.

The other key decision was adopting an LLM-first detection approach for non-English languages. Our English pipeline uses over 70 deterministic regex patterns built up over months of refinement. Translating those patterns idiomatically into every new language would have been impractical and fragile. Instead, non-English assessments route through a language-model-first detection pipeline supplemented by lightweight structural patterns. This approach is both more scalable and more linguistically accurate.

The result: adding a new language to the PAICE assessment pipeline now takes 20 to 30 hours of development effort. That is what makes the next section possible.

What's Next

Spanish, French, and Portuguese are the second wave. German, Dutch, and Danish are next. Our Q2 target is to support all major European languages (over 1-million native speakers). This allows us to provide PAICE well in advance of the August 1st effective date for the EU AI Act.

The goal is straightforward: any professional, anywhere, should be able to measure their AI collaboration capability in their own language. Assessment quality should not depend on which language you think in.

Want to assess your team's AI collaboration readiness across languages? Learn about PAICE for organizations or take an individual assessment to see it firsthand.

Get Involved:

Take the assessment (free, always, now in 5 languages)
Explore our Baseline offerings (for organizations)
Read the whitepaper (comprehensive framework)
Contact us about your specific requirements