BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Business Events Deutschland//DE
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
UID:33ef92739f12e647a4ffc0f5cf10404371ae1343@business.alle-events.de
DTSTAMP:20260908T163000Z
DTSTART:20260908T163000Z
DTEND:20260908T183000Z
SUMMARY:CAIML #44
DESCRIPTION:Cologne AI and Machine Learning Meetup CAIML #44 is going to 
 happen on September 8\, 2026\, at TH Köln. Thanks to TH Köln\, [Prof. Ge
 rnot Heisenberg](https://www.th-koeln.de/personen/gernot.heisenberg/)\, 
 and [KölnBusiness](https://koeln.business/) for their support!  **Agenda
 **  18h30 Open Doors 19h00 Welcome & Intro 19h15 [Tim Voßmerbäumer](http
 s://www.linkedin.com/in/tvossmer/) (Data & AI Consultant at Scalefree):T
 alk to Your Data: A Semantic-Layer Architecture for Reliable Analytics A
 gents  An LLM can produce valid SQL over a warehouse schema and still an
 swer the wrong question\, because the schema does not carry the organiza
 tional meaning of a metric. This talk presents an agentic system over Go
 ogleAds\, Google Analytics 4\, and Google Search Console data\, modeled 
 in Data Vault 2 on BigQuery\, in which a governed semantic layer serves 
 as the agent's knowledge interface. The agent answers along two paths: q
 uestions that map to a predefined metric go through the dbt Semantic Lay
 er\, using definitions that already exist rather than ones the model wri
 tes. Questions no predefined metric covers take an exploratory path\, wh
 ere the agent writes its own SQL under a rule-based guard. The agent may
  also decline when the data cannot support an answer. Without that optio
 n it does not fail visibly\, but produces SQL that passes every structur
 al check\, fabricatesthe rows\, and reports them faithfully.  \\- Short 
 Break \\-  19h50 [Ivan Herreros](https://www.linkedin.com/in/dr-ivan-her
 reros-b64a204/) \\(Senior AI Consultant \\| Machine Learning Engineer at
  inovex\\): From ChatGPT to Agentic AI: The Common Thread of AI Developm
 ent  How is Agentic AIactually progressing\, how will it keep reshaping 
 day-to-day work\, and is any of it really impossible to predict? To answ
 er that\, we trace the evolution of LLM-based applications: from natural
 -language dialogue to the delegation of digital workflows to agents that
  act autonomously over ever-longer time horizons. The common thread: vie
 wed with some distance\, most of the steps (from RAG through MCP to Clau
 de Code and Agent Harnesses) werenot just predictable but predicted. Dra
 wing on almost three years of building conversational and agentic AI pla
 tforms\, we show which patterns recur and what actually sits behind the 
 buzzwords. Whoever understands this underlying logic can situate new tec
 hnological developments proactively\, instead of reacting to every new h
 ype cycle.  20h20 Networking with food and drinks provided by KölnBusine
 ss  See you in September\, [Aaqib](https://www.linkedin.com/in/aaqib49/)
 \, [Marc](https://www.linkedin.com/in/marcmuellercs/) & [Fabian](https:/
 /www.linkedin.com/in/fabianhadiji/)\n\nQuelle: https://www.meetup.com/co
 logne-ai-and-machine-learning-meetup/events/313098107/
LOCATION:Deutschland\, Deutschland
URL:https://www.meetup.com/cologne-ai-and-machine-learning-meetup/events/
 313098107/
END:VEVENT
END:VCALENDAR