Skip to content
Decision Intelligence Enterprise decision platform powered by Drools
Intelligent Process Automation Enterprise workflow powered by Kogito, jBPM, and KIE Server
Modernization Migrate legacy systems to modern platforms
Comparison See how Aletyx compares with alternatives
Consulting Expert guidance from design to production
Training In-depth learning for Drools and Kogito
Documentation Technical guides and API references
Resources Blog posts, guides, and announcements
Community Join the Aletyx community
conferenceaidmn
January 15, 2025
3 min
AI_dev recording of Architecting the AI Bridge with LLMs

Watch: Architecting the AI Bridge with LLMs, DMN, and Drools at AI_dev Europe 2025

Watch our session at AI_dev to learn the enterprise AI architecture for decision intelligence. See the powerful combination of GenAI & Symbolic AI, with Drools & DMN.

In this tech talk, Tim Wuthenow and Alex Porcelli address a critical enterprise challenge: the fact that GenAI is statistically correct, but often situationally wrong. This lack of accuracy creates significant risks in high-stakes scenarios.

This tech talk covers a practical hybrid AI architecture that integrates the flexibility of LLMs with the precision of open-source standards like DMN and Drools. The talk demonstrated how this approach allows you to update AI reasoning in real-time, creating an intelligent system that is not only compliant but also instantly adaptable to changing business policies.

This session was presented on August 2025, at the AI_dev Summit held in Amsterdam.

Takeaways ✍️

  • Why GenAI Fails in Production
    Statistical correctness ≠ situational correctness. The talk explores the risks of relying solely on probabilistic models for critical decisions made by systems.
  • The Power of a Hybrid Approach
    How to use the Model Context Protocol (MCP) to seamlessly connect LLMs (GenAI, like Claude Code) to Symbolic AI that follows a deterministic approach. In this case, policies determined using DMN decisions, executed by Drools and made readily available for integration with Claude.
  • Instant Adaptation & Auditability
    This hybrid AI architecture allows you to modify the business policies and logic that guardrails AI execution in real-time, without model retraining. Additionally it maintains a full, explainable audit trail.
  • Live Demo – AI-Driven Clinical Decision Support (AI-CDS):
    A demonstration of an AI-enhanced solution that shows the difference between traditional GenAI usage compared to the combined guardrails of Symbolic AI. An AI CDS provides smarter, faster, and personalized recommendations for better decision-making by healthcare practitioners and patients.

About the Speakers 🎤

Alex Porcelli: A seasoned architect with over 25 years of experience. A long-time open-source advocate and contributor to Drools, jBPM, and Kogito for over 15 years.

Tim Wuthenow: An enterprise AI architect with 15+ years in business automation at IBM and Red Hat. Specialized in integrating rule engines with AI for compliant, explainable systems.

Resources & Links

🔗 Get the Slides: Architecting the AI Bridge, by Aletyx at AI_Dev Europe 2025 (Download).

🤩 Learn more about Aletyx build of Drools.

🚀 Ask for a Demo: See how our enterprise-grade Drools, Kogito, and DMN solutions, backed by built-in AI, can help your business scale GenAI with confidence. Get in touch, we’ll be happy to expand on your ideas.

Couldn’t make it to AI_dev Europe? Explore other events and tech talks, and get to know about upcoming events here.