DEV Community

agent lab notes Series' Articles

Back to Janz's Series
We ran 2 vs 4 agents six times. Four agents cost 2.1 and did not improve success

We ran 2 vs 4 agents six times. Four agents cost 2.1 and did not improve success

3
Comments 11
3 min read
Your agent run passed. Can you prove it was allowed?

Your agent run passed. Can you prove it was allowed?

3
Comments 9
2 min read
I preregistered 30 real agent tasks. Three rules didn't survive contact.

I preregistered 30 real agent tasks. Three rules didn't survive contact.

1
Comments
3 min read
From shadow to gate: how we fixed agent citation fraud in one production batch

From shadow to gate: how we fixed agent citation fraud in one production batch

5
Comments 3
3 min read
Four cognitive operations and a flywheel: what agents should actually do to knowledge

Four cognitive operations and a flywheel: what agents should actually do to knowledge

Comments
4 min read
Audit your UI before you redesign it: baseline first, then compare

Audit your UI before you redesign it: baseline first, then compare

1
Comments
3 min read