Jev is an AI model that doesn't write text. It only makes decisions, and it's fast and cheap.
In 10 minutes, @alexdotgs will show you what it does and answer the big question: "Where would I actually use this?"
Unlearn more 👇
Characterisation tests record what your code does today (not what you want it to do).
They're temporary and a bit ugly, but they'll quickly catch an agent changing behaviour mid-refactor when your normal suite won't.
Your agent can write them for you in a couple of minutes. We
In the age of agentic engineering, refactoring is disappearing from git history. GitClear's latest report puts moved lines at 3.8% of code changes so far this year, down from 21% in 2022.
Agents write a lot of code, but they don't go back and tidy up unless you ask them to.
Your tests can stay green while your app breaks. Ask an agent to tidy up a file with no rules in place, and it'll happily change behaviour.
We show exactly this at the start of Workflow 7 (Refactoring with an AI Agent), then spend the rest of the workflow making sure it doesn't
A one-line bug report, no repro steps. You guess, poke around, hope it sticks.
Half the time it doesn't, because you never proved you had the actual bug.
Workflow 06: Debugging and Performance fixes that.
Live now on Unlearn - learn more: unlearn.dev