What do agents discover?
Follow which documentation and tools they find, load, and choose while working toward a goal.
03 / AGENT EXPERIENCE
See your product through
an agent’s workflow.
Follow how agents discover and use your product. Inspect documentation reads, tool calls, errors, and workarounds in recorded runs, and connect friction findings to their evidence.
YOU BRING
Your product, realistic user goals, and the tools agents can access.
YOU COMPARE
Instructions, documentation, toolsets, and interface versions.
YOU LEAVE WITH
Evidence of recurring frictions and the changes worth investigating.
THE QUESTIONS THAT MATTER
Follow the actions behind the outcome.
Where does the experience break?
Inspect misunderstood instructions, unsuccessful tool use, recovery attempts, and repeated friction.
Did they use your product?
Look beyond a passing outcome to see whether the agent used your tools or reached the answer another way.
START WITH A BRIEF
Tell your agent
what you want to learn.
It can help turn your context into tasks, rubrics, and an experiment you can run again. Review the proposed tasks and criteria before launch.
EXAMPLE PROMPT
Check whether agents actually use our vector-database SDK or reach the answer another way. Inspect failed requests, retries, and fallback implementations, including SQLite. Separate correct answers from successful product use. Link each finding to trajectory evidence, propose a docs or tool change, and evaluate it against the current version.
HOW IT WORKS
From the first question
to your next change.
Give agents realistic goals
Set tasks that reflect customer work, with the interfaces and starting context they would actually receive.
Follow the work
Run trials across the relevant agent and model configurations. Inspect recorded trajectories, outputs, and friction findings.
Turn friction into a change
Group recurring problems, inspect supporting runs, and evaluate changes to docs, tools, or instructions.
EXPLORE THE PLATFORM
Your next eval starts
with your coding agent.
Give it the framework and infrastructure
to design, run, and analyze your evaluations.