Most tools that score your AI visibility hand you a number and no reasoning. I wanted one that shows its work.
Put in a URL and it fetches the page, reads it the way a crawler would, and runs twenty checks across crawler access, machine-readable content, structured data, and technical hygiene.
Every check is listed with what it's worth and whether it passed. Same page in, same score out, every time.
Built it on a Cloudflare Worker. It's also running on my agency site.
Every model has its own idea of whether a business is worth quoting. This checks whether a site gives them enough to work with. It isn't a perfect science and I keep updating it as the models change.
Questions about how it's built? Email me.
Ask Greg is a chatbot on my home page. Ask it about my background and it answers from what I’ve actually done.
Your question goes through a Cloudflare Worker to a Make.com workflow, and OpenAI writes the reply. I get an email for every conversation.
It also remembers. Come back on the same browser with the same company name and it picks up where you left off. One company never sees another company’s conversation.
I built it with Claude and ChatGPT doing a lot of the coding. Deciding what it should do, testing it and fixing what broke was on me.
Try it on the home page →
Moe does masonry in St. Louis. His site runs on a hosted website builder, which limits what you can change under the hood, so a lot of the usual technical SEO was off the table. I fixed a canonical bug that was keeping his interior pages from getting indexed on their own. Added FAQPage, LocalBusiness, and Service schema so search and answer engines could tell what he actually does.
Wrote a guide for people trying to figure out whether they need tuckpointing, repointing, or a full rebuild. Then I built him a chatbot. It answers from his own site content, never quotes a price, and sends anyone asking about estimates or scheduling to his phone. That was on purpose. A bot guessing at masonry prices would cost him money.
He's been a client for over ten years. Google keeps changing what it rewards, so the work never really stops — I'm tweaking the site, and when traffic dips I'll run something local while I wait to see if a fix took. A change can take weeks to get recrawled, and it might not work. Knowing which thing to try and how long to give it is the part you only get from doing it for a while.
Fitness data is scattered across apps that don't talk to each other. Your watch knows your sleep. A different app knows what you ate.
Nothing connects them. I'm building one place where they do.
Food, caffeine, supplements, workouts, movement, and mood in a single log, with AI reading the patterns and suggesting what to adjust.
Still in development.
A 3D trail game for young kids. They walk a path, collect berries, and stop for short learning moments: counting berries, counting the animals they passed, simple adding.
It reads everything out loud and never rushes them. I built it in Replit with AI doing a lot of the coding. Deciding what to teach, testing every trail and fixing what broke was on me. It plays in the browser now.
An app version is next. Still in testing.
This is a personal project and isn’t released for schools or commercial use yet.