Software Engineering Daily

Agentic Mesh with Eric Broda

49 min

Eric Broda, co-author of "Agentic Mesh," discusses the critical shift of AI agents from individual tools to large-scale, distributed enterprise ecosystems. The episode highlights the architectural challenges of deploying multi-agent systems in production, emphasizing the need for enterprise-grade capabilities like trust, explainability, and robust distributed computing principles. Broda advocates for an "Agentic Mesh" approach, where agents become full participants in business processes, transforming organizational agility and employee efficiency.

Summarized by Podsumo

🎧 Listen 🎙️ Ask about this episode

✨ Key Takeaways

1

Enterprise-Grade Challenges

Deploying AI agents in production requires addressing critical enterprise capabilities such as security, observability, traceability, and especially trust and explainability, which are often overlooked in lab environments.
2

Distributed Computing for Scale

With the expectation of thousands to millions of agents in enterprises, managing and orchestrating them becomes a distributed computing challenge, necessitating robust communication mechanisms like event streaming (e.g., Kafka) and architectural principles akin to microservices.
3

"Know Your Agent" (KYA) Framework

A foundational concept for building trust in autonomous agents, drawing parallels with human resource practices (Know Your Employee) and proposing a federated certification mechanism similar to Underwriters Laboratory.
4

Agent Explainability Gap

A key missing piece is the ability to log an agent's internal thought process and task plan *before* execution, allowing comparison with actual outcomes for better testing, debugging, and understanding of agent behavior.
5

Future of Work with Agents

Enterprises should leverage agents to augment existing employees, making them 10x more efficient and enabling new market opportunities, rather than solely focusing on cost reduction, to retain intellectual property and drive innovation.

💬 Notable Quotes

"The problem is, is despite the best efforts of the folks who are writing the agent, more often than not, there are folks that are new, perhaps, to the game, they're doing some proof of concepts. Or we have some very senior data scientists who may have outstanding capabilities. Most of the data science scientists submit our geniuses, but they don't have the understanding of the enterprise grade requirements to get something into production."
"If an agent is gonna do some of that work, we have to think really, really hard about how we're going to set up a trust framework so that we can actually trust the agents as well as we can trust the employees."
"If you used agents to get rid of people, you are losing the intellectual property that has built your company. That's a really, really big loss and I can't emphasize that more."