← All work

Our product

Kataster

The sales engine JT Digital built for itself reads the websites of 59,551 Slovenian companies on servers we run and drafts first emails that a person approves.

2026

59,551

Slovenian companies in the knowledge base, each website read by our crawler

~103,000

Lines of TypeScript, about 48,000 of them tests

478

Finished emails from the first version, in 16 sessions run by two people

In our first audit of live output, the language model had made up 27 of 53 facts about the companies it wrote to. It copied its own example sentence and called a casino a law firm. Now every fact has to share a number, a name or a word with the company's own website, or it is dropped.

Kataster's company page for a made-up brush workshop, scrolled to its Profile card, where the profile is marked NEEDS_ATTENTION with the note too_few_facts:3/6 and the Classification card below says a READY profile comes first.
Illustration Only three of the six facts the model wrote could be found on the company's own website, so Kataster parked the profile for a person and no email will be drafted from it. Drawn from the app's own screen with made-up sample data.

Want something like this?

Tell us what you need and we build a working prototype first, so you see it before you pay for it.

Tell us what you need built
← All work