Independent research

Research that stays close to real systems.

Arvexo studies how AI products get evaluated, understood and improved in the environments where people actually use them, not in isolated benchmarks.

1Publications so far
3Active questions
Follow on GitHubFounder profile
Q1AI effectiveness

Can AI adoption be measured without reducing people to productivity scores?

Methods for connecting real AI usage to outcomes that matter, without turning teams into dashboards.

Q2Knowledge systems

How do organizations find and spread the AI practices that actually work?

Studying how effective habits move from one person's workflow to an entire team's, and what breaks that transfer.

Q3Human-centered AI

What makes a system understandable enough to safely improve?

Interfaces and feedback loops that keep people able to teach, correct and trust the systems they rely on.

Notes

Adoption isn't the same as value

Login counts are the easiest AI metric to collect and the least useful one to act on.

Get in touch

Have a dataset, a case or a question worth exploring together?

Write in with what you're seeing, we read every message and follow up on the ones with a real problem behind them.Write to Arvexo