ML & AI Meetup at inpeek

Agenda

• 17:45 Open Door bei inpeek AG in Bern, Bogenschützenstrasse 9, 3008 Bern im 5. Stock, https://inpeek.ch/
• 18:00 Welcome
• 18:05 Talk 1: Fabian Hüni (inpeek) How We Built a Knowledge Graph–Powered RAG Chatbot for Structured Data
• 18:30 Talk 2: Simon Martinelli: Spec-driven Development: Wie KI alles (und nichts) verändert
• 19:00 Apéro
• 19:45 Closing

There is only limited capacity - so first come, first serve. Please be fair - if you cannot make, cancel your reservation.

Location and beverages are sponsored by inpeek AG - thanks a lot for supporting the AI and ML community in Bern!

About Talk1: "How We Built a Knowledge Graph–Powered RAG Chatbot for Structured Data" by Fabian Hüni

Abstract
Our product Summarix helps manage and prepare employee CVs, project references, and skills for public tenders and proposals. To make it easier to identify suitable employees or projects, we aimed to go beyond traditional filters and full-text search by enabling natural language queries. Unlike typical RAG approaches that rely on vector embeddings, our data is stored in a relational database (SAP HANA) and includes not only text but also numeric and rich relational information.
In this talk, we will present how we built our chatbot Summy using a Knowledge Graph for data representation, Large Language Models for SPARQL query generation, and an Agentic AI workflow that allows dynamic interactions while maintaining controlled autonomy. We will also explore how we established observability and an evaluation pipeline using the open-source tool Langfuse. Finally, we will share insights from our development process and key lessons learned.
This project was part of SAP’s Early Adopter Care Program for the SAP HANA Knowledge Graph Engine and was featured as a Success Story on SAP’s website.

Speaker
Fabian Hüni began his career at BKW with an apprenticeship as an IT specialist before earning his bachelor’s degree in Computer Science and pursuing a master’s in Artificial Intelligence in Medicine at the University of Bern. As an AI Engineer at inpeek, he leads the company’s AI Unit, combining his experience as a full-stack developer with theoretical expertise in artificial intelligence to bridge the gap between research and practical application.

About Talk2: "Spec-driven Development: Wie KI alles (und nichts) verändert" by Simon Martinelli

Abstract
KI beschleunigt die Softwareentwicklung, aber wir haben immer noch das alte Problem: Code wird zur Quelle der Wahrheit.

Was wäre, wenn wir das umdrehen würden?

Dieser Vortrag zeigt eine neue (alte) Art der Softwareentwicklung. Anforderungen beschreiben und alles Weitere – Diagramme, Modelle, Code, Tests – wird von KI generiert. Ändern sich die Anforderungen, wird alles automatisch aktualisiert. Schluss mit veralteten Dokumenten.

Sie erfahren, wie sie Anforderungen formulieren, die KI versteht, und wie sie alles – von den Geschäftsanforderungen bis zum funktionierenden Code – miteinander verknüpfen. Anhand eines realen Projekts zeige ich, wie das in der Praxis funktioniert.

Es geht nicht darum, Entwickler zu ersetzen. Es geht darum, KI die langweiligen Aufgaben übernehmen zu lassen, damit wir uns auf das Wesentliche konzentrieren können: die Geschäftsanforderungen zu verstehen.