appliedAI Developers Meetup @ Munich
The era of AI demos is over. The Munich AI Nexus is a senior-level meetup series for architects and engineers solving real-world production AI challenges.
At the Kickoff: During the 'From Hack to Scale' session, we asked: 'What’s actually breaking AI in production?'
The results were clear:
42% said 'context' is the most overlooked architectural layer,
41% flagged 'data quality and chaos' as their biggest blocker.
MEETUP FOCUS:
Theme: "Context – From Data Chaos to Clarity"
This module focuses on transforming raw, fragmented data into structured, trustworthy context for AI systems.You can't automate what you don't understand. Agents without clarity are just more expensive hallucinations.
Upvote the topic-related issues you find the most interesting.
Read on here to discover the 9 themes for our upcoming meetups and the final scope of our Production Readiness Blueprint.
SUBMIT YOUR CASE:
Share your toughest "Data & Context" challenge (even rough!)
Examples:
Keeping customer context consistent across dozens of systems (CRM, ERP, ticketing)
Enforcing GDPR/PII compliance in multi-tenant RAG without killing personalization
Debugging and auditing contextual reasoning so compliance teams trust outputs
AGENDA
18:00 - 18:05 - Welcome
18:05 - 18:15 - Recap & Opening
Survey recap: 42 % flagged context as the most overlooked layer; 41 % cited data chaos as the top blocker.
Focus for tonight: moving from hacky demos to scalable, trustworthy AI.
18:15 – 18:45 - Talk 1: Daniel Töws, codecentric AG - LLMs vs. Messy Data – Lessons from the Trenches
Chaotic CSVs → structured data: 30 K attributes, iterative refinement, and Pydantic.
Agentic Data Buddy: navigating 7 K tables to answer “Why aren’t these pants selling?” — balancing speed vs. trust.
Open question: Can we trust LLM-generated analysis at scale?
18:45 – 19:15 - Talk 2: Alexander K, Horsch - plot-RAG (pRAG): Visualizing & Optimizing RAG Performance – Smarter RAG Evaluation
Visualizing retrieval bottlenecks to optimize speed and accuracy.
Synthetic datasets that rival manual evaluation — GPTMB 2025 Best Paper Award.
Formula: defining the optimal retrieval window for performance vs. precision.
19:15 - 19:40 - Startup Challenge: Gauthier Salavert, AskUnali - Hybrid RAG
Scaling retrieval and context-engineering: hybrid retrieval, re-ranking, and deduplication strategies. The group helps solve it.
19:40 - 19:55 - Case Study: When Context Fails
Groundhog Day with AI customer service - 8+ escalation attempts, zero understanding.
Pattern analysis + alternative approaches.
19:55 – 21:30 - Closing & Networking
Host: Asaad Almutareb [founder of artiquare, appliedAI Developers MUC Meetup Curator]
The appliedAI Developers bring together engineers, researchers and tech leaders solving real-world AI challenges through knowledge exchange and collaboration.
Codecentric AG develops software solutions for the future, something everyone of our 500+ employees in each of our 13 locations in Germany is passionate about.
HOW TO GET THERE?
Codecentric Address: Google Maps.
By Car: In the underground parking garage, you will find a large Codecentric logo on the wall at the end of the parking level on either the 1st or 2nd basement floor. Please ring the doorbell at the entrance next to it and take the elevator to the 1st floor, where our reception is located.
By S-Bahn:
Take any S-Bahn to Ostbahnhof. In the lower level of the station, follow the signs towards Werksviertel/Friedenstraße. At the end, take the stairs up and cross the street at the zebra crossing. Then turn left and take the first right into Haager Straße. Walk straight for 300 meters, and you will see Codecentric PLAZA on the left. The entrance is at August-Everding-Str. 20. Take the elevator to the 1st floor and ring the doorbell.
Parking: Link .
The parking garage address: Hanne-Hiob-Straße 10, 81671 Munich