Education voucher at hand? Step into the fast lane: Contact us
Contact us
CANA - Compliance & AI-Act Navigation Assistant
CANA (Compliance & AI-act Navigation Assistant) is an AI-powered compliance assistant that helps small and medium-sized businesses understand and act on their obligations under the EU AI Act (Regulation (EU) 2024/1689). Most SMEs know the regulation applies to them but lack the time, legal expertise, and budget to interpret hundreds of pages of legal text. CANA closes that gap.
The product starts with an adaptive compliance check: a guided, multi-step questionnaire about the company, its AI systems, and how they are used. Behind it runs a deterministic, rule-based assessment engine — not a black-box model — so the same inputs always produce the same, auditable result. From the answers, CANA classifies the AI system's risk level (including prohibited practices and high-risk / Annex III cases), calculates a readiness score, and generates a prioritized, actionable task checklist that companies can track from open to done.
Every obligation can be explained on demand. A retrieval-augmented (RAG) assistant answers EU AI Act questions and explains each task with citations to the official regulation, so users always see the source rather than an unverified claim. A reference-database browser lets users explore the underlying articles directly, and finished plans can be exported as a printable PDF for internal documentation or audit preparation.
CANA is bilingual (English and German) and built on React 19 (TanStack Start) with a FastAPI/Python backend, a Groq LLM, and Supabase with pgvector for retrieval. It provides informational guidance, not legal advice.
The goal of CANA is to make EU AI Act compliance accessible to small and medium-sized businesses that lack dedicated legal or compliance teams. Rather than another document to read, CANA turns the regulation into a guided, adaptive assessment that classifies a company's AI systems, measures readiness, and produces a concrete, prioritized action plan. It grounds every explanation in citations to the official regulation, so users can trust and verify the guidance. Ultimately, CANA aims to lower the barrier to responsible AI adoption — helping SMEs move from uncertainty to a clear, auditable path toward compliance.
The Team
- T. Hai Nguyen
T. Hai Nguyen — AI & Data Quality Lead. Hai owned data quality and evaluation for CANA. He built the golden dataset used to validate the compliance rule engine, designed the testing and evaluation workflows, and made sure the assistant's answers stayed accurate, deterministic, and trustworthy. His focus was turning "it seems to work" into measurable, repeatable quality — the foundation that lets a compliance tool be relied upon rather than just demoed.
SocialsLinkedIn - Fabian Kaltenecker
Fabian Kaltenecker — AI Solution Architect. Fabian designed and built CANA end to end, across backend and frontend. He implemented the RAG pipeline, integrated the LLM for source-grounded answers, and built the adaptive assessment workflow that translates the EU AI Act into a guided check and task plan. He connected retrieval, vector search, and a deterministic rule engine into one coherent product.
SocialsLinkedIn - Roland Elsäßer
Roland Elsäßer — Project & Compliance Lead. Roland led project management and the regulatory backbone of CANA. He conducted the EU AI Act and market analysis, defined the compliance logic and audit process, and kept the team aligned on scope and delivery. He bridged legal requirements and product decisions, ensuring CANA reflects the regulation accurately and solves a real problem for SMEs
SocialsLinkedIn
We AI-proof
your career
