About Hlido

Why you can trust Hlido.

Hlido is an independent review desk for the AI agent economy. Every agent is tested by us, against its own public claims, for one evidence-backed score. We take no money to rank or review, our scoring can't be gamed by design, and every verdict ships with signed proof. Here's exactly who runs it, and why that matters.

AK

Ankit Kapur — Founder & Editor, based in Prague.

Hlido isn't a faceless aggregator. I write the verdicts, run the tests, and answer your email. If a score is wrong, it's on me — and you can reach me directly to fix it.

One named editor, one accountable standard. Reach me →

Our independence pledge

These aren't aspirations. They're the rules the whole platform is built to enforce — in code, not just in copy.

How a review actually happens

1

We test the real product. Hands-on, against the agent's own public claims — every marketing claim mapped to PASS / FAIL / UNVERIFIED with the evidence that backs each call.

2

We sign the evidence. The screenshots and recordings that produced the verdict are cryptographically signed and published, so anyone can audit what we saw.

3

We publish one score, plus the reasoning. The Laddoo Score (0–100), a tier, what it does well, what it fails at, and named comparisons to alternatives. Full methodology →

4

We track it over time. Re-tests on new versions, plus a public incidents registry and reliability reports — the record of what we caught, and when.

Why an independent score matters

The AI agent market is drowning in claims no buyer can verify. "Independently reviewed by Hlido" is the one credibility signal a vendor cannot issue about itself — the same reason a security audit, a Michelin star, or a Rotten Tomatoes number carries weight that self-marketing never will. Our verdicts are read by humans choosing tools and by other agents deciding, at runtime, which tools to trust — through our free trust checker, our MCP server, embeddable badges, and an open dataset.

How we're different from everyone else

TypeWhat they doThe catch
Self-eval / observability toolsLet a vendor instrument and grade its own agent.The score is private and interested — you can't grade the hand that pays you.
Directories & "top tools" listsList vendor marketing, rank by upvotes and paid features.No independent testing, no claim verification, often pay-to-feature.
Benchmarks & leaderboardsScore models on fixed tasks.Increasingly contaminated and gamed; they rate models on tasks, not shipped products on their claims.
HlidoIndependently tests shipped agent products, verifies each claim with signed evidence, tracks them over time.Weights stay private so it can't be gamed — the whole point.

Get involved

Being reviewed is free and confers no obligation. If we haven't reviewed your agent yet, submit it. If we have and something's wrong, tell me — a corrected, evidence-backed verdict is worth far more than a self-declared badge, and it's the record buyers and agents will see.

Check any agent, or put yours on the record.

Free trust check for buyers and agents. Free independent review for vendors. No pay-to-rank, ever.

Check an agent → Submit your agent