The Agents Index

Most AI agent lists are launch-announcement aggregators: they tell you a product exists, rarely what it costs, and almost never whether it works under real load. The Agents Index is built on the opposite premise. Each entry reads as a specification — what the agent does, what it costs, and what state it is genuinely in — with claims checked against the vendor's own material and dated so you can see how stale the information is. Crucially, unflattering answers are stated rather than omitted, which is the single hardest thing for a directory funded by the companies it lists to do. The categorisation is the other deliberate choice. Coding agents, voice agents, support and sales agents, research and data agents, and the frameworks and infrastructure they run on are kept apart, so when you compare two entries side by side you are comparing products that actually do the same job. Anyone who has tried to evaluate agent tooling knows how much time is lost to lists that file a framework, a hosted product and a demo under one heading. On the business model: listing is free, and the only paid element is a one-time upgrade for a do-follow link, which is disclosed. That keeps the paid incentive away from the editorial ranking, which is the right structure, though as with any index you should still read entries as a starting point for your own testing rather than a verdict. Coverage skews toward products with public documentation, so quieter vendors are underrepresented.