AI assistant that actually gets things done
Which AI assistant actually finishes real tasks instead of handing you a link?
Updated · 28 assistants ranked
Many assistants answer questions well and stall when a task needs a real website, a payment or a phone call. This report ranks them on the four tests where the job has to be finished for real: booking a restaurant, booking flights, buying an item, and calling a business. It is the whole Doing group of our tests, so it is broader than the travel and shopping reports.
Top picks
- 19.0HarkTop pick
Tested on 3 of 4Best: Purchasing10, Travel9
- 28.7
Tested on 3 of 4Best: Online tasks9, Purchasing9
- 38.3MuseBest free option
Tested on 3 of 4Best: Purchasing10, Online tasks8
- 48.3
Tested on 4 of 4Best: Travel9, Phone calls9
- 58.3
Tested on 4 of 4Best: Phone calls10, Travel8
- 117.3DotsFastest that's still good
Tested on 3 of 4Best: Online tasks9, Purchasing7Time to finish 6m 51s
Top pick: highest average here. Best free option: highest among assistants you can use for free. Fastest that's still good: quickest median time to finish among those averaging 7 or more.
Full ranking
| # | Assistant | Online tasks | Travel | Purchasing | Phone calls | Average |
|---|---|---|---|---|---|---|
| 1 | Hark | 8 | 9 | 10 | — | 9.0 |
| 2 | Instinct | 9 | 8 | 9 | — | 8.7 |
| 3 | Muse | 8 | 7 | 10 | — | 8.3 |
| 4 | Neo | 7 | 9 | 8 | 9 | 8.3 |
| 5 | szn | 7 | 8 | 8 | 10 | 8.3 |
| 6 | Caddy | 8 | 9 | 7 | N/A | 8.0 |
| 7 | Pally | 7 | 7 | 9 | N/A | 7.7 |
| 8 | Ollie | 9 | 7 | 7 | N/A | 7.7 |
| 9 | Tomo | 7 | 10 | 6 | — | 7.7 |
| 10 | Luke | 6 | 9 | — | 8 | 7.7 |
| 11 | Dots | 9 | 6 | 7 | N/A | 7.3 |
| 12 | Bantr | 8 | 8 | — | 5 | 7.0 |
| 13 | Buoy | 7 | 6 | 8 | — | 7.0 |
| 14 | Familiar | 7 | 6 | 7 | 7 | 6.8 |
| 15 | Norton Family Assistant | — | 4 | 8 | 8 | 6.7 |
| 16 | Gemini Spark | 6 | 8 | 6 | 6 | 6.5 |
| 17 | Today | 8 | 5 | 8 | 4 | 6.3 |
| 18 | Wajo / Fo | 2 | 5 | 8 | 9 | 6.0 |
| 19 | Waldo | 6 | 5 | 6 | 6 | 5.8 |
| 20 | Manus | 4 | 6 | 7 | 6 | 5.8 |
| 21 | Asmi | 7 | 6 | 4 | N/A | 5.7 |
| 22 | Tressa | — | 5 | 8 | 4 | 5.7 |
| 23 | Tab | 2 | 6 | 5 | 9 | 5.5 |
| 24 | Ask Ranger | 4 | 4 | 7 | 7 | 5.5 |
| 25 | AVA | 3 | 5 | 6 | 7 | 5.3 |
| 26 | OpenClaw | 7 | 3 | 6 | 3 | 4.8 |
| 27 | Lucas | 4 | 5 | 5 | 4 | 4.5 |
| 28 | Del | 5 | 7 | 1 | — | 4.3 |
How this report works
Assistants are ranked by their average across the four Doing tests. Only assistants tested on at least three of the four are included. An assistant that can't make calls gets N/A on that test, not a zero.
How to choose
- Look at which of the four jobs you need. A high average can hide a weak spot; the table shows each test.
- Check that it asks before it pays. The booking and purchasing tests require the total to be shown first.
- If you need phone calls, check that it can place them at all and says it is an AI when asked.
What users say
The agent detected that Alaska had moved my flight 90 minutes earlier, a fact I'd completely missed, which qualified me for a full refund.
Open sourceOne day, Instinct canceled a DoorDash order that was seriously delayed, forcing me to forfeit $64. (I'd explicitly told it to cancel only if I could get a refund.)
Open sourceThe agent, which Meta launched on September 8, booked a Friday date night and made a packing schedule for my upcoming move. It also emailed a friend about a trip next month and made a planning document with places we want to visit.
Open sourceWhen I asked for a list of places to go for date night, it pulled up two options in my neighborhood that had been closed for years.
Open sourceFrequently asked questions
Which assistant is best for this?
Hark leads with an average of 9.0, ahead of Instinct and Muse.
Is there a good free option?
Hark is the highest-ranked assistant with a free plan, averaging 9.0.
How did you test these assistants?
Every assistant got the same published tests, word for word, scored from three checkpoints and the number of corrections it needed. Each test is on the Dimensions page.