The easiest agents to try will not tell you what they cost
25 of the 84 agents in this directory publish no price anywhere, and they are not spread evenly. Of the 25 you can start using by texting a number, 13 say they are free, 9 say nothing at all, and 3 publish a paid price. Behind a signup it reverses: 23 of 28 publish one.
The 25 you can just text
No account, no app, no waitlist. You send a message and it works, and this is what each one says it costs.
It reverses the moment you have to sign up
Every listing, grouped by what it asks of you first, against whether it publishes a price at all. Both fields are read off the product’s own pages.
| How you start | Publish a price | Who |
|---|---|---|
| Phone call | 0 of 1 | Boardy |
| Invite only | 1 of 4 | Instinct, Mae, Wajo, Wis.ai |
| Install an app | 1 of 3 | 222, Alfi, Orbits |
| Not stated | 1 of 2 | Miso, Rene |
| Text a number | 16 of 25 | Arlo, Beside, Boba, Caddy, DUO, Dunnit, Eugene, Flip, Homer, Iris, Jarvie, Juke, Mochi, Navi, Orchid, Ori, Predicts, Series, Served, Soar, Tomo, Zeme, asaply, remi, textmyagent |
| Waitlist | 2 of 3 | Air by WZRD, Matcha, Tab |
| Add a number to a group | 3 of 4 | Ollie, Shuffle, Sidekicks, brobot |
| Tap one link | 9 of 11 | Ember, Kachi, Nudge, Otty, Pally, Stanley, Super, Tempo, Zapia, folk, iChatWithGPT |
| Create an account | 23 of 28 | Airtap, Arlo (arlomind.com), Bo, Boski, Brea, Catch, Ditto, Duckbill, Espa, Howie, Lindy, Lucas, Manoa, Martin, Maxi, Muse, Notis, Persona, Poke, Sabi, Slashy, Town, TryOpenClaw, Vellum, Zo Computer, alfred_, allora, iHermes |
| Self-host it | 3 of 3 | Hermes, OpenClaw, OpenInstinct |
The pattern is not that cheap things hide their price. It is that the products designed to be tried on impulse have nothing to gain from answering the question before you are in the thread, and the ones that need an account have already decided to ask you for something. Whichever way round it works, the effect on you is the same: the fastest thing to try is the one you can least easily budget for.
Breadth is about the job, not the price
The obvious next question is whether paying more gets you a product that does more. Across every listing that publishes a price, the rank correlation between price and the number of capabilities it states is -0.03, which is nothing. What breadth actually tracks is what the thing is for: the 13 social and introductions state 3.0 capabilities on average, against 6.3 across the 54 personal assistants. A narrow product is narrow because of the job, not because of what it costs.