AI Agents Hub
About AI Agents Hub
An independent, hand-curated directory of AI agents worth knowing about.
Why this exists
The AI agent landscape is moving fast — too fast for anyone to keep track of by hand. Existing directories are either overwhelming (thousands of entries, no signal) or promotional (every tool is "the best"). This site aims for the middle ground: a tight, opinionated list of 62 agents with honest notes on each.
How we pick
Every agent here has to clear a bar:
- Real traction — active development and a genuine user base.
- Clear use case — you should know what it's for in one sentence.
- Accessible — you can actually try it, not just watch a demo.
- Distinct — it does something the others in its category don't.
What the badges mean
Open Source — code is public and self-hostable. Commercial — paid product, often closed-source. Freemium — free tier with paid upgrades.
Independence
This is an independent project. Listings are not paid placements. If that ever changes, it will be disclosed clearly on the listing itself.
Suggest an agent
Spotted an agent we're missing? Found something outdated? The cleanest way is to open an issue or pull request on the source repository. Every listing is just a data file — easy to add or fix.
AI Agents Hub is an independent directory. All product names, logos, and brands are property of their respective owners.
Editorial team
AI Agents Hub is run by a small editorial team that has been working on AI tooling and developer infrastructure since 2019. Every entry on this site has been written, edited, and signed off by a person — not generated, not scraped, not translated by machine.
Mavis has spent the last six years writing about developer tools and AI infrastructure. Before AI Agents Hub, she worked on the docs team at a major LLM provider and shipped two open-source developer tools of her own. She personally tests every agent listed on this site against the four-point bar (real traction, clear use case, accessible, distinct) and writes the long-form analysis posts on the blog. When she is wrong about a tool, the tool wins — and the directory is updated to reflect what users actually report.
Anders runs the technical-review pass for every agent that lands in the directory. He verifies the benchmarks, the pricing claims, and the SWE-bench / GAIA / Humanity's Last Exam numbers we cite. He also catches the "this worked in 2024 but the API changed" kind of rot that aggregators miss.
Editorial process
- Submission or discovery — an agent comes in via email, a reader tip, or our own monitoring of GitHub trending and product launches.
- First review — the editor tests the agent against the four-point bar. If it fails any of them, the submission is rejected with a written reason.
- Drafting the entry — the editor writes the description, the verdict, and the relevant comparison data from scratch. No copy-paste from the tool's marketing page.
- Technical review — Anders verifies the claims, runs the benchmarks where they matter, and fact-checks the pricing.
- Publication & upkeep — the entry goes live. We re-review every entry on a 90-day rotation and update on reader reports.
Corrections policy
If you find an error — wrong pricing, stale benchmark, misattributed quote — the fastest fix is to email us with a link to the source. Verified corrections go live within 24 hours and the entry's last-updated date is bumped. We do not delete old content quietly; every major change is logged.