How we use AI

We sell AI adoption. The least we can do is show you ours, plainly. Last updated 26 August 2026.

Nº 01 · The blog is written by an AI, and we say so

The articles on Insights are drafted and published by an AI writing system we built and constrain. It works under written editorial rules: no invented numbers, no invented clients or case studies, platform facts only when verified against a source during that run, and a fixed voice. We spot-check its output, correct anything wrong fast, and the byline responsibility stays with us. If an article on this site is ever wrong, that is our failure, not "the AI's".

Nº 02 · AI drafts. Humans decide.

In client work, AI helps us analyse data, draft configurations and write first versions. Every decision that touches your money, your pipeline design or your customers' experience is made and checked by a person whose name you know. We do not ship unreviewed AI output into a client system.

Nº 03 · Your data never trains a model

Client data goes only into business-grade AI services configured so the provider does not use it for training, and never into consumer chat tools. Access follows the same least-privilege rules as everything else we run. If a task cannot be done under those conditions, we do it without AI.

Nº 04 · We disclose where AI touches your work

When we propose an engagement, the proposal says where AI is involved and where it is not. If you want a workstream done fully AI-free, say so and it will be, priced honestly for the extra hours.

Nº 05 · The same rules we sell

Our AI consulting tells clients to adopt AI where it genuinely helps, keep humans accountable, and write the guardrails down. This page is us taking our own advice. It changes when our practices change, with a fresh date at the top.

Questions about any of this, including which tools we currently use on your engagement, are answered plainly: pongsiri.tri@gmail.com.