Upgrade to Pro — share decks privately, control downloads, hide ads and more …

BiPRO-Tag 2025: Pragmatische Gen AI - Smarte Ab...

BiPRO-Tag 2025: Pragmatische Gen AI - Smarte Abläufe & echte Integration

Generative AI ist kein Selbstzweck – es geht darum, echte Aufgaben zu lösen und Systeme intelligent zu unterstützen.
In diesem Vortrag zeigt Christian Weyer anhand pragmatischer Muster, wie Small & Large Language Models und andere AI-Komponenten sinnvoll zusammenarbeiten können: für smarte Suchen, clevere Entscheidungen und strukturierte Ergebnisse. Statt Hype und Bauchgefühl gibt es hier konkrete Ansätze, wie Gen AI sicher und transparent in bestehende Anwendungen und Geschäftslogik integriert werden kann – live und hands-on.

Avatar for Christian Weyer

Christian Weyer

June 04, 2025
Tweet

More Decks by Christian Weyer

Other Decks in Programming

Transcript

  1. § Technology catalyst § AI-powered solutions § Pragmatic end-to-end architectures

    § Microsoft MVP for AI § Google GDE for Web AI § [email protected] § https://www.thinktecture.com Pragmatische Gen AI Smarte Abläufe & echte Integration Christian Weyer Co-Founder & CTO @ Thinktecture AG
  2. Pragmatische Gen AI Smarte Abläufe & echte Integration Our journey

    3 Intelligent Interfaces Decision Intelligence Conversational AI Document Intelligence AI Model Integration
  3. Language Models understand and generate semantically rich human language, transforming

    it into text or structured data for both humans and machines. ⚠ Non-deterministic: same input can lead to different outputs. Embedding Models capture semantic meaning by encoding human language into numerical vector representations, facilitating understanding, comparison, and retrieval for both humans and machines. ✅ Deterministic: same input always results in the same embedding. Pragmatische Gen AI Smarte Abläufe & echte Integration 5 🫱 🫲 Generative AI
  4. § Language & embedding models part of end-to-end architectures §

    Embedding models can be run locally § Optimized for CPU § Language models still hard to run locally § High GPU power § High VRAM § High memory bandwidth Pragmatische Gen AI Smarte Abläufe & echte Integration API-based AI model integrations 6
  5. Pragmatische Gen AI Smarte Abläufe & echte Integration Classical business

    applications & UIs 8 API-based data Document-based data
  6. Pragmatische Gen AI Smarte Abläufe & echte Integration Talking to

    documents (Retrieval-augmented generation) Cleanup & Split Text Embedding Question Text Embedding Save Query Relevant Results Question Answ er w / sources LLM Embedding Model Embedding Model 💡 Indexing / Embedding Question Answering .md, .docx, .pdf etc. “What should I do…?” Vector DB 9
  7. Pragmatische Gen AI Smarte Abläufe & echte Integration Talking to

    APIs (Function / Tool calling) 10 “When is CW available for a two-days workshop?” System Prompt (+ employee data) + Schema (for structured output) Web API Availability business logic
  8. Pragmatische Gen AI Smarte Abläufe & echte Integration DEMO 11

    § Searching for company policies § Querying expert availability
  9. Pragmatische Gen AI Smarte Abläufe & echte Integration Semantics-based decisions

    Guarding (e.g. prompt injection) Routing (selecting correct target) “Lorem ipsum…?” Target Knowledge Base / RAG Target Internal Systems / API Call Target … something else … Fine-tuned Language Model Embedding Model 13
  10. Pragmatische Gen AI Smarte Abläufe & echte Integration DEMO Recording

    damage assessment in a conversational way 19
  11. Pragmatische Gen AI Smarte Abläufe & echte Integration YOUR journey

    24 Intelligent Interfaces Decision Intelligence Conversational AI Document Intelligence AI Model Integration