AI agents evolve beyond simple chatbots by using large language models, instructions, and tools to integrate directly into business processes.
Azure AI Foundry provides a comprehensive platform to address common developer challenges like model selection, security, and observability when building AI agents.
The Agent Service offers enterprise-grade features including orchestration, SDK integrations, knowledge tools, and built-in content safety for robust agent development.
The AI application lifecycle consists of ideation, implementation, and operations, starting with low-cost experimentation in GitHub Models before moving to Azure Foundry.
Use GitHub Models to rapidly prototype by comparing different prompts and models against test datasets and using evaluators to generate performance KPIs.
Implement a robust evaluation strategy by incorporating KPI checks into CI/CD pipelines and using continuous monitoring with end-to-end tracing in production.
A practical example demonstrates a multi-agent system that analyzes contracts, checks compliance, and uses automated evaluations for continuous quality assurance.
Design effective multi-agent systems by selecting the appropriate interaction pattern, such as sequential, concurrent, or group chat, based on your process and outcome goals.
Use the Semantic Kernel agent and process frameworks to implement complex multi-agent workflows and deploy them scalably across various environments.
23:33Oliver Will
26:54Marcel Scherenberg
24:22Philipp Schmid
28:53Mike
27:07Eric Jadi & Rene Pajta
39:11Thomas Dohmke
23:39Dennis Zielke & Rene Pajta
29:03Jörg Neumann
Jobs that call for the skills explored in this talk.
Commerz Direktservice GmbH
Duisburg, Germany
Devoteam
adesso SE
Deutsches Medizinrechenzentrum GmbH