> Markdown version of [/videos/1538-composable-intelligence-how-henkel-and-microsoft-are-shaping-the-agent-ecosystem?t=772](https://www.wearedevelopers.com/videos/1538-composable-intelligence-how-henkel-and-microsoft-are-shaping-the-agent-ecosystem?t=772). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Composable Intelligence: How Henkel and Microsoft Are Shaping the Agent Ecosystem

Standard LLMs fail at complex enterprise tasks. Learn how Henkel and Microsoft engineered a composable, semi-autonomous agent ecosystem that achieves governed autonomy without stifling developer innovation.

- **Speakers:** [Dennis Zielke](https://www.wearedevelopers.com/@dennis-zielke), [Manuel Schettler](https://www.wearedevelopers.com/@manuel-schettler)
- **Event:** World Congress 2025
- **Published:** August 20, 2025
- **Duration:** 27:18
- **URL:** https://www.wearedevelopers.com/videos/1538-composable-intelligence-how-henkel-and-microsoft-are-shaping-the-agent-ecosystem

## Summary

Henkel, balancing a massive product catalog with scattered institutional knowledge, partners with Microsoft to transition from simple chatbots to sophisticated AI agents. Because standard language models lack the domain specificity required for complex chemical application queries, the organizations developed an agentic ecosystem capable of semi-autonomous decision-making, API integration, and contextual memory. This shift allows lab technicians and sales representatives to use hands-free, voice-based interactions to accurately compare product properties and application conditions.

By adopting a composable, three-layer architecture, engineering teams achieve "governed autonomy"—providing guardrails without stifling developer innovation. Leveraging open standards like the Model Context Protocol (MCP) and agent-to-agent (O2A) communication alongside OAuth 2.0 and OIDC, the system securely connects structured relational databases with unstructured SharePoint repositories. Developers integrate OpenTelemetry to continuously monitor multi-step model inferences and ensure reliable fallbacks if an API fails during a live prompt sequence.

Deploying enterprise AI reveals that traditional retrieval-augmented generation (RAG) is insufficient for complex data relationships; instead, intelligent systems must combine knowledge graphs with persistent, user-specific session memory. Testing these non-deterministic agents requires continuous validation frameworks that extend far beyond traditional software release cycles. As enterprises navigate the transition to agent-first interfaces, they must manage user expectations, recognizing that "first cars looked like carriages" while gently pushing users past legacy chatbot paradigms toward truly proactive assistance.

**Keywords:** henkel AI adoption, AI agent ecosystems, model context protocol MCP, agent-to-agent communication, O2A protocol, governed autonomy framework, enterprise AI composability, domain-specific LLM deployment, knowledge graph integration, contextual AI memory, RAG limitations, voice-based AI assistants, LLM open telemetry, non-deterministic AI testing, proactive agent behavior

## Chapters

1. **Introduction to building real-world AI agent solutions** (00:04) — An overview of the partnership focusing on building and applying agentic solutions to real-world industrial problems.
1. **Scaling product knowledge with proactive enterprise voice agents** (00:45) — How a vast product catalog and scattered knowledge bases create the need for highly context-aware voice agents in chemical research.
1. **Defining true agent behavior versus traditional enterprise chatbots** (05:48) — Why modern agents must act semi-autonomously to decompose objectives, plan ahead, and invoke specialized APIs in complex scenarios.
1. **Balancing technology capabilities with market demands in innovation** (07:37) — A collaborative framework for safely evaluating and pushing new technological capabilities toward impactful production business solutions.
1. **Designing an open agent platform with governed autonomy** (10:03) — Establishing platform principles that give developers freedom to innovate while maintaining guardrails against compliance and budget pitfalls.
1. **Architectural building blocks for enterprise agent development platforms** (12:52) — How to layer essential features like workflow logic, data access, memory, observability, and strong developer experience above the cloud infrastructure.
1. **Overcoming data complexity and context retrieval formatting challenges** (17:02) — Using knowledge graphs and memory to bridge structured relational databases with unstructured data for nuanced, context-aware prompt generation.
1. **Implementing open standards for distributed agent communication** (18:48) — Leveraging open protocols to ensure secure data sharing, agent discovery, and bidirectional agent-to-agent interactions at an enterprise scale.
1. **Applying resilient software principles to slow generative models** (20:44) — Migrating twelve-factor app principles and robust error handling directly into prompts to compensate for unpredictable API responses and slow models.
1. **Demonstrating a voice-assisted agent for hands-free workflows** (22:32) — A practical exercise showing how a voice-driven agent queries product catalogs and manages complex constraints without a traditional user interface.
1. **Addressing data quality challenges and organizational adoption barriers** (25:56) — Why providing exceptional data quality and managing user resistance to unfamiliar agent interfaces remain the hardest challenges in adoption.

## Related Moments

- [Designing agentic AI solutions for the enterprise](https://www.wearedevelopers.com/videos/1831-ai-for-enterprise-developers-dr-damir-dobric) (from "AI for Enterprise Developers - Dr. Damir Dobric")
- [Leveraging the comprehensive generative artificial intelligence stack](https://www.wearedevelopers.com/videos/969-make-it-simple-using-generative-ai-to-accelerate-learning) (from "Make it simple, using generative AI to accelerate learning")
- [Shifting focus from isolated models to enterprise AI systems](https://www.wearedevelopers.com/videos/100130-ai-in-production-applied-ai-enterprise-use-cases) (from "AI in Production: applied AI & enterprise use cases")
- [Understanding language models and autonomous executing agents](https://www.wearedevelopers.com/videos/1725-wearedevelopers-live-build-a-multi-ai-agents-game-master-with-strands-our-weekly-web-finds) (from "WeAreDevelopers LIVE - Build a multi AI agents game master with Strands & our weekly web finds")
- [Building autonomous functions with conversational agent frameworks](https://www.wearedevelopers.com/videos/1624-30-powerful-aws-hacks-in-just-30-minutes-boost-your-developer-productivity) (from "30 powerful AWS hacks in just 30 minutes: Boost your developer productivity")
- [Summarizing developer experience and artificial intelligence companions](https://www.wearedevelopers.com/videos/884-forget-developer-platforms-think-developer-productivity) (from "Forget Developer Platforms, Think Developer Productivity!")

## Related Articles

- [What is Agentic Programming and Why Should Developers Care?](https://www.wearedevelopers.com/magazine/625-what-is-agentic-programming-and-why-should-developers-care)
- [Stephan Gillich - Bringing AI Everywhere](https://www.wearedevelopers.com/magazine/489-stephan-gillich-bringing-ai-everywhere)
- [WWC24 Talk - Scott Hanselman - AI: Superhero or Supervillain?](https://www.wearedevelopers.com/magazine/469-wwc24-talk-scott-hanselman-ai-superhero-or-supervillain)
- [Graph and AI Trends 2026: Why Is AI Running but Not Yet Delivering?](https://www.wearedevelopers.com/magazine/680-graph-and-ai-trends-2026-why-is-ai-running-but-not-yet-delivering)

## Related Jobs

- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [Senior AI Agent Software Engineer (Go, Python) (m/f/x)](https://www.wearedevelopers.com/jobs/48277-senior-ai-agent-software-engineer-go-python-m-f-x) at **Dynatrace**
- [Principal Field Architect - AI Agents](https://www.wearedevelopers.com/jobs/ext/1442858-principal-field-architect-ai-agents) at **Twilio**
- [Head of AI Applications](https://www.wearedevelopers.com/jobs/ext/1456210-head-of-ai-applications) at **ZEISS Group**
- [Senior Backend Developer — AI: MCP & Agent Engine](https://www.wearedevelopers.com/jobs/48297-senior-backend-developer-ai-mcp-agent-engine) at **basebox GmbH**
- [Head of AI Applications](https://www.wearedevelopers.com/jobs/ext/1231536-head-of-ai-applications) at **ZEISS Group**