Material for people who evaluate
Measurements, architecture and the security model — written for someone who has to judge whether this holds up, not for someone browsing. Each of the three says what it covers, so you can skip the two that are not your question.
Guides on introducing AI in a company
These exist, in depth, and they are in German — twelve pieces written for the DACH market on what an introduction costs, why pilots stall before production, how works councils are involved, and what the EU AI Act expects of a company using such systems. A short English section on the Regulation exists as well — linked below, because the dates changed in July 2026 and most summaries in either language still show the old table.
- Introducing AI in a company de — the whole path from first use case to running system.
- What an introduction costs de — the items that quotes tend to leave out.
- Why pilots do not reach production de — the four reasons, and which of them is fatal.
- EU AI Act: scope, dates, penalties en — four pages, written for companies outside the EU. Annex III moved to December 2027; Article 50 did not move at all.
- All guides de
What is available
Benchmark Results
ARGUS performance on standard multi-hop QA datasets: MuSiQue, HotpotQA, and 2WikiMultiHopQA. Tested on 500 samples per benchmark.
See the numbersArchitecture Overview
How graph-based retrieval, personalised PageRank and multi-step reasoning let a smaller open-weight model reach an answer that brute-force retrieval misses — which is what makes running inference on our own hardware affordable.
Explore the engineSecurity & Compliance
On-premise deployment, data sovereignty, and EU AI Act compliance — how ARGUS is designed for regulated industries.
Learn about securityBenchmarked, Not Promised
Want the Full Technical Brief?
Schedule a call and we'll walk you through the architecture, benchmarks, and deployment options for your specific use case.
Talk to Us