Can an AI agent get the job done on your site?
We send a Claude agent to do a real task on your website, the way a customer's assistant would, and score the run
with AXIS , Netlify's open Agent Experience Index.
You see every step it took and exactly where it got stuck.
How the score works
AXIS scores a run on four things. Goal (40%): a judge grades the task against a weighted rubric, checking the agent's claims
against what the site actually showed. Service (20%): did your site respond reliably, with controls an agent can find by name?
Agent (20%): were the agent's decisions sound, right-sized and necessary? Environment (20%): did the browser harness itself behave?
Process scores combine success, speed and efficiency, weighted so failures count more than successes, then map onto a 0 to 100 curve.
Scoring is a faithful port of AXIS by Netlify (MIT licence).
Every report includes the scenario as AXIS JSON, so you can rerun the same test in AXIS against other agents.