Find out what a change broke, before a customer does
Save the questions that matter as a test suite. Run it after a change, and each case passes or fails against the agent's own model and script. Nothing is dialled, and nobody is called.
On every plan- No call placed
- A pass or a fail on each
- You fix what failed
Every fix can break something else
A small edit, a new mistake
You improve the pricing answer, and the refund answer quietly gets worse. Nobody finds out until a customer does.
Testing by hand gets skipped
Ringing your own agent with the same ten questions after each edit is dull. So it stops happening.
A bad call teaches nothing
A call went wrong last week. Unless it becomes a check, the same mistake can come back.
From a question to a pass or a fail
Write a case
Type what the caller says, and how to check the reply: it must say something, must not say something, or is judged against what you describe.
Or take it from a real call
Turn a real call into a case. The caller's words play in order, while the agent answers live.
Run the suite
Each case goes to the agent's own model, with the prompt a real call would use.
Read the result
Every case shows pass or fail, and the reply it got. Fix what failed, and run it again.
A suite, after a run
Which cases passed, which failed, and what the agent really said.
- A case from a real call replays the caller's words, and the check reads the whole replayed conversation.
- A run tests the words. It does not reproduce the voice, interruptions, tools or the knowledge base.
- For a hard customer who argues back, use the Simulation Lab. A test asks set questions; a rehearsal holds a conversation.
The reply it got sits under each case. Nothing was dialled.
Run suiteWhat you set
An agent or a team
A team's suite runs through its coordinator, so routing and handovers are tested too. Teams need a plan with agent teams.
The check
Contains, does not contain, or a judged check for answers that can be worded many ways.
The size of a suite
Up to 50 cases and 120 caller messages a run. Split a bigger set into several suites.
When it runs
Only when you press Run. Nothing runs by itself.
Who it helps
Whoever edits the agent
A quick answer to whether the last change broke anything.
Operations heads
A check to ask for before a changed agent goes back in front of customers.
Which plan has it, and what it costs
| Plan | Starter₹999 a month | Growth₹2,999 a month | Professional₹7,999 a month | Enterprise₹28,000 a month |
|---|---|---|---|---|
| Agent tests | Included | Included | Included | Included |
| Tests for agent teams | Not included | Included | Included | Included |
What it costs
- ₹0.88 per test run. Each run of a suite that actually ran. A case counts for nothing on its own: the charge is per run.
Free trial: Included, at no charge, with a daily limit on runs.
See every plan and rateWatch out for this
A pass says the words were right. It does not test the voice, interruptions, tools or the knowledge base, so make a real test call before you go live.
It works with these
Agent tests: questions
Does a test ring anyone?
How is this different from the Simulation Lab?
Can I test what went wrong on a real call?
Does it use my knowledge base?
Save your first test.
Agent tests are on every plan, from the free trial.