# Fractional CTO Notes on Building AI Products

Long-form field notes on safeguards, resumability, explainability, model strategy, and product taste.

- **Canonical URL:** https://vidhata.me/notes
- **Author:** Vidhatanand V. (Vid)
- **Role:** Fractional CTO and AI Systems Architect

- [Models may judge. Code must guarantee.](https://vidhata.me/notes/models-may-judge-code-must-guarantee) - Use models for ambiguity. Use deterministic software for permissions, money, safety, state changes, and irreversible actions.
- [A workflow that cannot resume is still a demo.](https://vidhata.me/notes/a-workflow-that-cannot-resume-is-still-a-demo) - Persist state, checkpoints, cancellation, recovery, and first-class artefacts before adding another prompt revision.
- [Every recommendation deserves a why.](https://vidhata.me/notes/every-recommendation-deserves-a-why) - Preserve the source, confidence, and reasoning a person needs to understand, challenge, and improve the result.
- [The best model is rarely the moat.](https://vidhata.me/notes/the-best-model-is-rarely-the-moat) - The edge lives in context, data, workflow, evaluation, tools, trust, distribution, and the speed of the learning loop.
- [Taste is part of correctness.](https://vidhata.me/notes/taste-is-part-of-correctness) - A passing test is not enough when the interface is confusing, the narrative is lifeless, or the output fails the human bar.

## Work with Vid

Start with the actual technical pressure: [bring the problem](https://vidhata.me/hire).