Silicon Friendly

Ben Gendler, Michael Vavilov, Christian Onochie, Lakshay Gupta, Shubham, and Siarhei Palishchuk built something that basically no one realized they needed until now: a way to measure how well your website works with AI agents. Silicon Friendly scores sites from L0 (basically useless to an AI) to L5 (the AI could live there and do your job for you). They've already rated 1,381 websites and the results are kind of terrifying—only 13 sites in the world have reached L5, and 259 of them are stuck at L0. That's a lot of companies spending millions on web infrastructure that AI can't even read.

Here's why this matters: every company is racing to build AI agents, but nobody's asking whether their website can actually handle them. Silicon Friendly runs 30 automated checks across five levels—can agents read your content (L1), find what they're looking for (L2), actually talk to your forms and APIs (L3), integrate and perform actions (L4), or operate autonomously on your site like a digital employee (L5). The big dogs are all over L5—Zapier, Postman, DeepL, Auth0, Sanity, Plaid, PayPal, Square. But the vast majority of the internet is still stuck in L2 and below, which means their fancy AI integrations are going to hit a wall the second the agent actually tries to do something.

What I love about this is it's solving a problem nobody was talking about six months ago. Everyone's building AI products, but nobody benchmarked whether the web itself was ready. Silicon Friendly just became the Yelp for AI-readiness, and if you're a developer or product team not checking your score, you're basically flying blind. They're based somewhere in the AI infrastructure space and they're probably looking for more sites to grade and companies who want to know where they stand.