Skip to content

Projects

We love what we build

We develop websites, web applications and digital automations. Either as a custom software product for companies or as a web app for anyone who wants to offer their users something unique.

A gallery wall of screen designs, a visitor standing in front of it.
konerio on a tablet, next to notebooks and a pen.

konerio

konerio, the brand-new learning platform for academies or businesses that want to offer courses to staff, partners and customers. A broad feature set, a cool design and easy handling, both for the provider and for the learners. Hosted in the EU, under your own brand and custom domain. Two weeks' free trial with the full range of features.

The GESIBA project map on a screen in an office.

GESIBA

Project managers regularly need maps showing the sites of several group companies. The system processes the addresses entered, assigns each company its colour and generates a labelled map for print within seconds.

The AHR Rechtsanwälte website on a phone, held in a car.

ahr-recht

Together. For your rights – AHR Rechtsanwälte in Kooperation, a Vienna law firm, get a well-considered website with an editing system and an informative blog. When we design websites for law firms, we make sure clients find the area of law and the responsible lawyer without detours.

The trainCo website on a phone, held in a metro carriage.

trainCo

A lively web presence for the trainco education academy. Professional and in-house training, hands-on and up to date, carried by a confident design and current technology (Next.js, automated course dates, booking platform).

Software of our own

Our own software runs on the same technical foundation as our client projects. It is also where we test new developments in practice. So our client projects always run on the hottest sh…, er, state of the art.

jukeep in the browser: the library of knowledge collections.

jukeep

Makes company data searchable. Answers come exclusively from your own documents – each with a precise source reference.

yoorda in the browser: the home page with the ordering app on a phone.

yoorda

Ordering and checkout system for hospitality. Menus are captured digitally, read out and translated automatically.

neurit in the browser: the home page with the question field.

neurit

An analysis tool that examines a business question from several angles to prepare a decision.

The best of our repos

From the webse lab

  • 01

    TUNADOJO — Fine-tuning small language models for a specific purpose

    Large AI models know a little about everything. For specialised tasks a much smaller model is often enough, adapted to one subject area (fine-tuning via LoRA). We test precisely at which model size a task is still solved reliably and where the model starts guessing. The upside: the AI runs on our own hardware, data stays in house, and there are no recurring subscription fees.

  • 02

    KONERIO — Learning systems with strict tenant separation

    When several companies use the same software, their data has to stay strictly separate – even if the application has a bug. That is why we anchor the protection in the database itself (row-level security): without a valid tenant key the database simply returns no rows. We test whether the barrier holds with deliberately faulty and manipulated queries.

  • 03

    UMWIDMUNGSRADAR — Detecting changes in official sources automatically

    Laws and standards change without anyone being notified. Our system fetches official sources at regular intervals, compares the current state with the previous one and puts changed paragraphs side by side. The key lies in telling them apart: the system only raises an alarm for genuine changes in substance, not for reformatting.

  • 04

    EUDAMED — Comparing public EU data over time

    Public EU data on tenders and funding only becomes meaningful over a span of years. Because formats and identifiers have changed several times across decades, we merge these data sets into continuous time series. The real work is in the preparation: getting mismatched years into a technical shape that makes them comparable at all.

  • 05

    RAGFROG — Building a vector search of our own

    A vector search finds text by meaning rather than by exact terms (embeddings). To understand what actually matters technically, we built a search of our own. The most important finding: it is not the search algorithm that decides whether results are good, but how cleanly the texts are split up beforehand (chunking).

We, you and your software. Loyal for the long run.

We only build systems that we operate securely and keep current for you over many years.

A racing pit crew servicing a robotic lawnmower in a driveway.

Your project. We take it on.

Tell us what you need. We will say what makes technical sense — and what other agencies leave out.

Contact

webse.at · Mag. David Paci

Weiherweg 4, 9500 Villach, Austria

info@webse.at+43 (0)650 4352 452