5 places to find early-stage startups before they raise.

TL;DR

The most interesting startups leave public traces long before they raise money, in places that are open to anyone but rarely connected: the trade registry, the patent office, accelerator cohorts, university spin-off lists and grant databases.

Finding them takes two things. The search has to read those local sources alongside the global ones, and it has to start from the problem you want solved, described in plain language, because a keyword only finds the companies that happen to use your words.

The startups everyone can see

If a startup is on a conference stage or in a funding announcement, every scouting team in your industry has seen it too. The companies that fit a specific problem are often smaller and quieter than that: a team of twelve in a regional accelerator, or a university spin-off with one patent application and no press at all.

A funding round is usually the first moment a startup becomes visible to everyone, and by then the company has often been around for a year or two. It existed on paper the whole time, in places that are public but that few people search. These are the five we find most useful.

The trade registry

Every company is registered somewhere, and the record tells you when it was founded, what legal form it has, which activity it declared and, most usefully, whether it is still active. It is also the quickest way to see which companies are still active today.

The patent office

A patent application tells you what a team is building, often a year or more before there is a product page. It names the inventors, and it shows which technical problem they consider worth protecting. For hardware, materials and life sciences this is frequently the earliest public trace.

Accelerator and incubator cohorts

Programs publish their cohorts, and a place in a selective one is an early sign of quality. The focus of the program also tells you what the company does more reliably than a tag, since somebody who knows the field chose them for it.

University spin-off lists

Technology transfer offices list the companies that came out of their labs. If you scout in deep tech, these lists are worth more than any conference, and the companies on them are often years ahead of any press coverage.

Grant databases

National and European innovation grants are public. A grant means that independent experts read the plan and found it credible, and the award notice usually describes what the company promised to build. Hardly anyone scouts there, which is part of the appeal.

Why this rarely gets done

The obvious problem is that each of these sources is local, comes in its own format and often its own language, and links to none of the others. Doing this by hand for one country is a project, and doing it for thirty is a department.

Collecting the records is the easier half. The harder half is keeping the wrong ones out. Kuanta reads global and local sources side by side and holds 3.9M+ validated startups in 187 countries. A profile is only kept when the company is live, is under ten years old and is a startup, so law firms, consultancies, investors and twenty-year-old companies stay out.

Search for the problem, in your own words

The other half of the fix is the way you search. A keyword finds the companies that happen to use your words, whereas a sentence can describe the problem you want solved.

We recently searched for "pre-seed startups in Rotterdam which automatically clean ships with robots" across 3.9M+ startups and found 24 that match, ranked by fit. As a tag filter that search would not work, and it is exactly the kind of question a corporate scouting team actually has.

See it live at Valencia Digital Summit

Joep Weterman, our Head of Product, takes this further at Valencia Digital Summit on Thursday 22 October, in his session "Finding Signals in the Noise: Startup Search After the Agentic Shift". If you are in Valencia this week, come and find us.

Sources

Kuanta platform data, October 2026.