Public data is not a product
ÚPSVaR publishes employment agency licenses as Excel dumps. We turned three messy registries into a searchable product anyone can use in seconds. How Nextrey builds software from ugly real-world sources.
Most people assume that if something is published by the state it is also usable.
It is not.
In Slovakia employment agency licenses sit on ÚPSVaR. Three registries. Three separate pages. Excel files you download and open yourself. One for temporary employment agencies. One for paid job intermediaries. One for supported employment agencies. Different layouts. Different sheet names. Sometimes a name and an address jammed into a single cell. Sometimes a tab with a trailing space that breaks a naive parser.
If you need to check whether a company is actually allowed to operate you do not open a search box. You open a spreadsheet. You scroll. You hope the file is current. You hope you are looking at the right registry. You hope you did not miss a cancelled permit hiding on another sheet.
That is an archive. It is not a product.
Why this matters
For a company hiring through an agency the question is simple. Is this partner legitimate right now?
Not last year. Not according to their website. According to the official record.
When the official record is hard to use people skip the check. Or they check once poorly. Or they trust a PDF someone emailed them. That is how risk sneaks into a normal business process. Not through drama. Through friction.
The same pattern shows up everywhere. Public registries. Legacy exports. Vendor portals that only spit out files. Internal tools that “have the data” but make nobody faster. The information exists. The decision still takes too long.
What we built
At Nextrey we treat that gap as the real problem.
We built a service that pulls the official Excel registries every day. It parses all three sources. It stores active suspended and cancelled records. It normalizes the messy parts. It exposes a signed API for search by company name or company ID. On top of that sits a simple public checker so anyone can verify a license in seconds instead of hunting through downloads.
The interesting work was not the UI.
The interesting work was accepting the source as it is. Date named sheets. Inconsistent columns. Excel serial dates. Physical person rows where the address lives inside the name field. Sync jobs that must fail loudly when the file format quietly changes. Caching so the product feels instant. Alerts when the morning import breaks.
Ugly input. Clean output. That is the job.
You can try the checker here: https://www.personalneagentury.sk/overenie-licencie-agentury
What this says about how we work
We do not start from a blank Figma file and pretend the world is tidy.
We start from what users already suffer through. Downloads. Manual search. Split systems. Half finished admin panels. SaaS ideas that need roles billing and multi tenant boundaries on day one. Mobile apps that only make sense if the backend already exists. Old applications that still pay the bills and need modernization without a big bang rewrite. AI features that help only when they sit inside a real workflow not as a demo chat box on the side.
The stack changes. The standard does not.
Architecture that will survive first users. Delivery that reaches production. Launch that does not fall over. Ongoing build after the applause ends.
That is the work behind Nextrey. SaaS. Web. Mobile. Modernization. Automation and AI when the product actually needs it.
The point
Government Excel is an extreme example. The pattern is common.
If your users still download files to answer a yes or no question you do not have a data problem. You have a product problem.
Fix the product and the data finally becomes useful.
If you are sitting on a process like that and want it turned into something people will actually open we are easy to find.