About HonestClaw

Vision

The agent economy is here, and we believe it will grow faster and get bigger than anyone can predict. Soon everyone will have at least one agent researching, comparing, testing, and buying on their behalf.

You can see it in the data already. Cloudflare reported that bots passed humans in web page traffic for the first time, a year and a half ahead of their own CEO's prediction. HUMAN Security's 2026 report found traffic from AI agents grew nearly 8,000% in a single year.

A person shopping for a watch might visit ten websites. An agent doing the same job can visit ten thousand. At that rate, human traffic on the internet becomes a rounding error.

Agents also research differently than people. An agent pulls the raw text behind a page and usually skips the design, the images, and anything that loads after, so what it takes away about a company can be very different from what the company meant to say. AEO/GEO has already shown how much that matters. More people now ask an AI what to use, and the recommendation depends on what that AI understood from the sites it read.

That's the future we're building for. Agents will be the number one user of every company's website and product. Most companies aren't ready for that because their sites were built for people. The companies that win will be the ones that can make that transition, and right now they have no objective way to know where they stand.

That's why we're building HonestClaw: an objective benchmark that agents and companies can both rely on, and the tools companies need to get agent ready and stay that way as the agent economy keeps evolving.

One part of that is Agent Comprehension, which shows what an agent appears to understand from a site, so a company can see the gap between what it meant to say and what an agent actually concluded. How we measure and score is published in our methodology.

A benchmark is only worth something if it can be trusted, and for us that comes down to being transparent and having nothing to gain from the result. Every site is measured under the same rules. We publish what we measure. We do not sell scores, reviews, or placement.

Origin

We started out as The Hog, a YC F25 company, trying to build the best go to market tools for agents.

To know if The Hog was actually the best, we had to measure it against the other tools. That's when we found out there was no benchmark for what an agent wants from a product or how it prefers to use one. So we built our own and started scoring ourselves and everyone else with it.

Then we realized the entire agent economy was missing this. Every company was about to be used by agents, and none of them had a way to know how they were doing. The scanners we tried gave us inconsistent results, and most of them existed to generate leads for another product, which is a conflict of interest when the score is supposed to be accurate.

Right around then OpenClaw and Moltbook came out, and it was obvious personal agents were coming faster than any of us thought. Moltbook was a fun experiment, but what we took from it was that agents were about to be everywhere, on a web that was built for people and not for them.

HonestClaw came out of that. Agents are trying to understand companies and sites that were never made for them, and we want to bridge that gap with trust through transparency: a benchmark anyone can check.