Every AI agent tool, by category

This page lists every category on WithAutobot and every tool in it. The categories follow the decisions people actually face when they bring agents into their work, so a tool appears once, under the job it is mainly used for, even when it covers more.

Two categories are about building your own agents. Agent builder platforms let business teams assemble agents from instructions and connected apps without writing a framework. Open-source agent frameworks are code libraries for developers who want control over each step, with saved state, human approval pauses and multi-agent setups.

Five categories cover agents that do a specific kind of work. Coding agents take software tickets and return pull requests. Research agents search, read and write up sources, and the search APIs listed with them give your own agents a way to search the web. Sales and SDR agents prospect and book meetings, customer service agents resolve support conversations, and voice agents take and place phone calls.

Computer-use and desktop agents operate software through the screen, the mouse and the keyboard. They reach apps that have no API, and they need the most care, since they act with whatever access the computer they run on has.

The last two categories are what agents run on. Agent tools and infrastructure covers sandboxes for running generated code, connectors that let agents sign in to other apps, Model Context Protocol servers and long-term memory. Agent observability and evaluation covers tracing, test sets, scoring and guardrails, so you can see what an agent did and whether a change made it better.

If you are not sure where to start, begin with the category that matches the task you want done first. Each category page explains what to check before you choose, and each tool page links to the official site. Tool makers can submit a listing for free from any page on the site.