AI agent
Software that takes a goal, decides its own sequence of steps, and calls tools to carry them out.
The line between an agent and a script is that a script's steps are written by a person and an agent's are chosen at run time. That is the entire source of both the usefulness and the unpredictability.
In practice most useful agents are narrow: they do one job, with a small set of tools, on a defined kind of input. Broad autonomous agents are still the exception rather than the norm in production.
Because the steps are chosen rather than written, an agent has to be measured rather than read. That is why every agent here carries a success rate from actual runs.
Judging an agent from a demo. A demo is one sample from a distribution, and the variance between runs is often larger than the difference between two products.
Related terms
See AI agent on a real listing
Every term here shows up in the catalogue next to a real result, with the findings published and the limits stated. Free to browse, no account needed.
Open the catalogue