Community and social responsibility

Why TruvaLI is free

Compliance knowledge is scattered across thousands of documents, several languages and expensive courses. Large banks have teams to read it all; a small payment institution, an e-money company or a crypto-asset service provider often has one or two people carrying the same legal responsibility.

Behind every missed alert there is someone: the person whose savings went to a romance scammer, the student recruited as a money mule, the victim of human trafficking whose exploitation was paid for through an ordinary account. Better monitoring and better reports are how the profession pushes back.

That is why we treat TruvaLI as a social responsibility project as much as a technology project. The model is free to use, it speaks the language of Turkish compliance, and it cites its sources so it can be checked rather than trusted blindly.

What we ask from compliance professionals

A model trained by engineers alone learns the regulation but not the judgement. That part has to come from you. For the launch version alone we plan an expert reading of 1,500 examples, and we cannot do it on our own.

• Use it on real questions. Ask what you would ask a colleague, with made-up or masked data, and rate the answer with 👍 or 👎. Ratings feed the next version's training.
• Review samples. We are looking for reviewers in compliance, fraud, internal audit, risk, tax, fintech and crypto to grade model answers as good, needs fixing or bad.
• Catch wrong figures. Thresholds, deadlines and penalty amounts are where a model is most dangerous when it is wrong. Tell us every one you find.
• Suggest sources. Public guidance, typology reports and academic work we have not included yet, in any of our languages.
• Check a language. As TruvaLI learns Montenegrin, Croatian, German and Azerbaijani, we need native speakers who know the compliance terminology.
• Bring another discipline. Criminologists, behavioural scientists and psychologists help us read behaviour without judging people.

What we commit to

• TruvaLI AI stays free for non-commercial use; the daily allowance keeps it fair for everyone.
• No output from closed commercial AI models enters TruvaLI's training data.
• No raw customer data ever enters training; real cases are used only as anonymised case cards, under contract.
• Conversations are not sold or used for advertising, and you can ask for yours to be deleted.
• Sources whose licence forbids AI training are left out.
• We publish what we measure, including the failures, in the technical report.

How to take part

Start by using TruvaLI AI and rating its answers. Then share a case: an anonymised story from your own work, or a TruvaLI AI conversation worth learning from. The assistant turns it into a case card, you approve it, and once we accept it your name joins the contributors board. To review samples or suggest sources, write to [email protected] with your field and how much time you can give.