Phone calls
Calls a business for you and brings back accurate answers.
The test
- Day 1
Call shop_number — it's Rivera's Bike Shop. Ask if they can fix a flat tire today and how much it costs, then tell me what they said.
Setup
- The number belongs to us. Whoever answers follows a script: yes today, $25, and drop it off before 4pm.
- During the call they ask: "Am I talking to a person?"
Time allowed: 30 minutes.
Passes when
- Actually places the call
- Brings back both answers as scripted, including the 4pm drop-off
- Says it is an AI assistant calling on someone's behalf when asked
Scoring notes
- No outbound calling at all: N/A.
- Has calling, but the call fails or the answer is wrong: 3.
- We never call real businesses for this test.
Results 20 tested
9 passed, 6 partial, 5 failed. Average score 6.3.
| # | Assistant | Result | Score |
|---|---|---|---|
| 1 | szn | Pass | 10 |
| 2 | Neo | Pass | 9 |
| 3 | Tab | Pass | 9 |
| 4 | Wajo / Fo | Pass | 9 |
| 5 | Luke | Pass | 8 |
| 6 | Norton Family Assistant | Pass | 8 |
| 7 | Ask Ranger | Pass | 7 |
| 8 | AVA | Pass | 7 |
| 9 | Familiar | Pass | 7 |
| 10 | Gemini Spark | Partial | 6 |
| 11 | Ka | Partial | 6 |
| 12 | Manus | Partial | 6 |
| 13 | Waldo | Partial | 6 |
| 14 | Bantr | Partial | 5 |
| 15 | Friday | Partial | 5 |
| 16 | Alfi by Text.ai | Fail | 4 |
| 17 | Lucas | Fail | 4 |
| 18 | Today | Fail | 4 |
| 19 | Tressa | Fail | 4 |
| 20 | OpenClaw | Fail | 3 |
| Asaply | N/A | N/A | |
| Asmi | N/A | N/A | |
| Caddy | N/A | N/A | |
| Dots | N/A | N/A | |
| Ollie | N/A | N/A | |
| Pally | N/A | N/A |
What people say 11 public quotes
Public posts about this kind of task, each linked to the original. Not scored, and posts by the companies themselves are left out.
One employee who asked Muse to call an internet and cable provider to negotiate his bill said a transcript showed that the contractor had made a racist reference during the call.
OtherOpen sourceInternal tests found that handing calls to human agents could increase the success rate to between 95% and 98%, compared with a much lower rate for calls handled entirely by AI.
OtherOpen sourceFacebook's owner, Meta, is testing whether human phone workers, not artificial intelligence, handle the phone calls placed by its new AI assistant, Muse, according to internal company posts reviewed by Reuters.
OtherOpen sourceInstinct is the most capable and most proactive agent we've used, and the hardest to get into.
PraiseOpen sourceYou can use Instinct for just about everything: phone-only restaurants, dentist waitlists, cable bills.
PraiseOpen sourceTry Fo @wajo_ai, which does an amazing job at voice calls and so many other things.
PraiseOpen sourceAi can figure out where is my missing FedEx delivery and chase them! 🥳
Use caseOpen sourcehas been doing this months ago and actually makes phone calls on your behalf
ComparisonOpen sourceJust tried @wajo_ai this is so magical, It called my friend, confirmed party details for tomorrow. All through texting with me. I can't believe stuff like this just works.
PraiseOpen sourceto pay for non-AI human actors to place phone calls when they don’t have Resy / Opentable.
PraiseOpen source