Independent review Est. 2026
Nouvre Review

A review of AI that actually works.

The archive

Past issues

  1. The benchmark after the benchmark

    What remains when a model’s headline numbers meet an ordinary working day.

  2. A small model, properly placed

    Three narrow systems that succeed by knowing exactly where their work ends.

  3. The interface is the evaluation

    Why the design around a model often matters more than another point of accuracy.

  4. Working code: the document room

    A plain-language search tool built for records that resist tidy databases.

  5. Against the artificial colleague

    Software need not impersonate a person to become useful company.

About the review

A practical record.

Nouvre Review is an independent publication from Nouvre about artificial intelligence in practice: the research worth keeping, the code that holds up, and the companies building with care.

It is published periodically, when there is something useful to report. Each issue is written for readers who prefer evidence to spectacle.