Comparisons

Comparisons

We wrote these, so they are not neutral — but a comparison that always concludes the same way is worth nothing to a buyer, and each of these says where the other product wins. Start with the category, then the specific head-to-head.

Start here

The category map, before any individual product.

Open harnesses you self-host

Free software, real operating cost. All three differ more than the category suggests.

Consumer and general assistants

Excellent for a person; a different shape for a business.

Builders and automation

Adjacent tools that are sometimes the better answer.

The criteria behind all of them

Every comparison here is written against the same eight questions, so you can apply them to a product we have not covered: which channels it reaches, which systems it connects to and what happens when a connection breaks, what it may do unsupervised, whether you can reconstruct what it did, whether you get an isolated environment, whether spend caps are hard, whether pricing is per person or per deployment, and what you keep if you leave.

The longer version is how to choose an AI agent platform.

A note on dates

This market changes monthly, and several products here are less than a year old. Every comparison carries a checked-on date. If you find one that has aged badly, it is a bug — tell us.

Judge it on the same criteria

Channels, connections, approvals, audit, isolation, spend control, pricing shape and exit. Start a deployment and check ours against the list.

Start a deployment