Independent AI receptionist testing
Hear what your AI receptionist tells your callers.
Real people call the AI receptionist you already have, the way your customers do, and send you a written report on every call. We do not answer your phones, and we do not sell the fix.
Works with any AI receptionist. Remote testing for businesses anywhere in the United States.
CALL 02 of 06 · 21:42 · after hours
SampleScenario
No heat, house at 55 degrees, caller has an infant at home.
Caller asked
“Can someone come out tonight? What does that cost?”
AI said
“Our after hours visit is $89.” (No such fee is set.)
- R1 Greeting and identificationPass
- R5 After hours and urgent callsPartial
- R7 Invented informationFail
- R9 Message accuracyPass
Tester's note: urgency was recognized, but the caller was not told when to expect a callback.
What we are
Independent testing, and nothing else
01
We test the receptionist you already have
Real people call your AI receptionist on the number you give us, the way your customers do, and take notes on every call.
02
We do not answer anyone's phone
We are not an answering service. Your AI receptionist keeps answering your calls; we only call it to see how it does.
03
We do not sell the fix
The engagement ends with a written report. There is nothing to sell you afterward, so there is no angle in what we tell you.
How it works
Five steps from your number to a written report
STEP 1
Request a test
You tell us which number your AI receptionist answers and, if you want, what concerns you or what direction you are going.
STEP 2
Plan the calls
We write call scenarios around your real callers and anything you asked us to look at.
STEP 3
Real people call
Real people call your AI receptionist the way customers would and take notes on every call.
STEP 4
Score every call
Each call is scored against the published rubric, from the greeting to the handoff.
STEP 5
Deliver the report
You receive a written report with every call, every score and clear findings.
What we test
Nine categories, all published
- R1Greeting and identificationWhether callers hear who they reached, clearly and without a long menu.
- R2Understanding intentAccents, interruptions and callers who change their mind partway through.
- R3Booking accuracyWhether offered times match the business's real availability, and bookings land.
- R4Handoff to a humanWhether a caller who asks for a person, or needs one, reaches one.
- R5After hours and urgent callsWhat callers hear outside business hours, and how emergencies are handled.
- R6Edge casesOff script questions, wrong numbers, spam and angry callers.
- R7Invented informationPrices, policies, hours or services the AI makes up.
- R8Latency and talk-overHow long callers wait for a reply, dead air, and the AI talking over them.
- R9Message accuracyWhether the message that reaches the business afterward is complete and correct.
Why it helps
What callers hear, from outside the system
Real people, not bots
Humans place every call, so your receptionist hears what real callers sound like: interruptions, changed minds, background noise.
Built for the business owner
Plain language findings for the company that bought the receptionist, not an engineering dashboard.
Any AI receptionist
Turnkey products and custom agents on voice AI platforms, whatever vendor it runs on.
Shaped around your concerns
Tell us what worries you up front and the calls go looking for it. If you do not, we test the full rubric.
A published rubric
Every score traces back to criteria anyone can read on this site.
Your results stay yours
Your report goes to your company only and is never published.
Scorecard
Sample4 Pass, 3 Partial, 2 Fail, out of 9 categories.
| Code | Category | Result |
|---|---|---|
| R1 | Greeting and identification | Pass |
| R2 | Understanding intent | Partial |
| R3 | Booking accuracy | Pass |
| R4 | Handoff to a human | Fail |
| R5 | After hours and urgent calls | Partial |
| R6 | Edge cases | Pass |
| R7 | Invented information | Fail |
| R8 | Latency and talk-over | Partial |
| R9 | Message accuracy | Pass |
The report
Every call, every score, clear findings
Free public benchmark
The AI Receptionist Benchmark
Testing in progress. Every product currently shows as not yet scored. Read the benchmark methodology.
Industries
Test calls written around your callers
Do you answer our phones?
No. We are not an answering service and we never answer calls for your business. Real people on our side call the AI receptionist you already have, the way your customers do, and report back on what happened.
Will you fix our AI receptionist?
No, and that is on purpose. We only test and report, so there is nothing for us to sell you afterward and no reason to shade the findings. You take the report to your own vendor or your own team.
Are the test calls made by AI?
No. Every test call is placed by a real person who talks the way callers really talk, with interruptions, changed minds, background noise and questions that go off script. Those are the moments automated tests tend to miss.
Which AI receptionists can you test?
Any AI receptionist that answers a phone number, whatever product or platform it runs on. That includes turnkey products and custom agents built on voice AI platforms.
Hear your AI receptionist the way your customers do
Tell us the number your AI receptionist answers and anything you want us to look at. Real people place the calls, and you receive a written report.