First public testimonials, training underway: where TEAF stands today

Something new has been happening with TEAF over the past few weeks, and I'd rather say it directly than wait for a bigger, more formal recap: the first public testimonials about this side project are starting to appear. I won't call them proof here — more on that below, and the distinction matters — but after a year spent defending a deliberately anonymized track record, seeing feedback expressed publicly changes something in how I approach what comes next. Quite simply, I'm proud of it.

What's actually moving: training and certification

In parallel, several efforts are progressing to give the five roles TEAF defines — Architecture Owner, Decision Steward, AI Control Officer, Capability Owner, Compliance Liaison — a sturdier framework. Training paths and certifications are being finalized, with the goal that these roles stop being theoretical definitions in a book and become verifiable, assessable, transferable skills. I'm deliberately not sharing a date or detailed content at this stage: I'll announce a certification once its content is ready to be rigorously assessed, not before — the same reason I never announce a date for the method's own scientific validation.

Same advice as always: go through the official repo

Faced with this early momentum, one caution stays unchanged, and I'll repeat it deliberately: the recommended starting point for experimenting with TEAF remains the official TEAF Light GitLab repo — not an end-to-end reconstruction generated by an AI assistant from the books' PDFs alone, with no human oversight. I recently detailed on this site a scenario illustrating exactly that risk: a fully AI-driven rollout can technically close the TEAF loop successfully, while producing unacceptable drift the moment no human validates structural decisions anymore. That isn't a theoretical caution on my part — it's exactly the kind of approach I advise against, however impressive the results look on paper. An approach too heavily assisted by AI, without the five roles held by identified humans, isn't a shortcut to TEAF — it's a way of bypassing what TEAF is supposed to guarantee.

What's still missing, and where it's genuinely progressing

The most important point remains the one I already made in the article on what TEAF has proven in the field: field feedback, even backed by public testimonials, is not scientific validation, and I won't let anyone believe otherwise. What's genuinely changing, though, is how dense the available metrics have become. Data gathered across recent rollouts is more concrete and more systematic than it was a few months ago — less anecdotal, more comparable from one context to another. It still isn't the reproducible, published measurement protocol TEAF needs to be credible beyond qualitative field experience, but it's a real step in that direction, not one more statement of intent.

Why I'd rather say this as it happens

I could wait until there's a complete protocol, published training and more testimonials before revisiting this. I'd rather document progress as it happens, under the same rule that applies to everything else on this site: say what's moving without presenting it as settled, and say what's missing without downplaying it.

• First public testimonials on TEAF: a real development, and one I'm proud of — but field feedback stays field feedback, not proof.
• Training and certification for the five TEAF roles: being finalized, with no date announced before they're ready to be assessed.
• The official TEAF Light GitLab repo remains the recommended starting point for experimentation — not an AI-assisted reconstruction with no human oversight.
• Available metrics are genuinely getting denser, but the scientific validation protocol remains work in progress, not an already-settled result.