The Stack Overflow Podcast Treating your agents like microservices
49 snips
Dec 5, 2025 Guillaume de Saint-Marc, VP of Engineering at Outshift by Cisco, dives deep into the exciting world of multi-agent architectures and their future as microservices. He explains the shift from single LLMs to collaborating specialized agents tailored for enterprise needs. The importance of communication protocols, security challenges, and creating an interoperable ecosystem are covered. Guillaume also discusses innovative tools like Slim for secure group messaging and highlights the potential applications and limitations of agentic systems.
AI Snips
Chapters
Transcript
Episode notes
Reasoning Over Size
- LLMs showed both knowledge access and emerging reasoning, changing how we solve problems.
- Breaking problems into many inferences can outperform one giant model call by mimicking human stepwise reasoning.
Specialization Beats One-Size-Fits-All
- Agents mix deterministic code and probabilistic LLM behavior, creating new computer science challenges.
- Enterprises will trust highly specialized agents rather than a single universal agent for critical tasks.
Favor Open Interoperability
- Build agents with open interoperable protocols to avoid vendor lock-in and enable cross-platform collaboration.
- Use open source standards like Agency to let different vendors' agents interoperate securely.