A few of you asked what happened to TLD.is, so here’s the honest version.
TLD.is is an AI domain appraisal tool. I paused it to do a full audit, and this is what I found:
Accuracy: not perfect, but way closer than the big names.
My appraisals weren’t spot on. But I tested them against real sales on NameBio, and in most of my tests TLD.is came in far closer to the actual sale price than Atom, GoDaddy or EstiBot. On one name that really sold for $9,600, two of the big tools overshot by 3 to 6 times. They’re not even in the same ballpark. HumbleWorth was the only one that came out about even with mine.
The comparable sales weren’t always real.
I checked the reports against NameBio. Some of the comps the AI showed didn’t match any real sale, so that’s getting fixed first.
Free appraisals were leaking.
My API bill didn’t match my paying customers. Also, the free limit worked per network, not per person.
Credits were tied to the browser.
If you cleared your cookies or switched computers, your paid credits were gone.
Housekeeping.
I had extra API keys active from different developers, a few percent of requests erroring out, and site wording that didn’t match how the free domain appraisal actually worked.
What the rebuild will have:
- Comps from real sales data only, never AI-invented.
- User accounts, so your credits follow you.
- Cheaper pricing aimed at domain investors.
TLD.is redirects to veml.io for now. It’ll be back sooner rather than later, and I’ll post here when it relaunches.
Lessons for anyone building a domain tool:
- Use real sales data. Never let AI invent comps.
- Build accounts before you charge money.
- Keep one API key, and check your bill against your sales every week.
Questions welcome.
David