ClueCon Weekly with Diego Gosmar [Sn. 15 Ep. 22]: The Practical Path to “Agentic AI

ClueCon Weekly with Diego Gosmar [Sn. 15 Ep. 22]: The Practical Path to “Agentic AI

In this episode of ClueCon Weekly, we’re joined by Diego Gosmar, open source telephony veteran (FreeSWITCH/Asterisk) and AI leader active across Linux Foundation communities—to break down what “agentic AI” actually means in practice.

We start with a reality check: sometimes the best solution isn’t an LLM at all. Diego shares real examples where classic ML and computer vision outperform multimodal models. Then we dig into where LLMs do shine, and why “agents” are more than just a fancy label for workflows.
​
Together, we cover:
▪️What separates an agent from a workflow (memory, actions, and decision-making)
▪️Why multi-agent systems are showing up in real production applications
▪️How teams use multiple agents to improve accuracy, reduce hallucinations, and manage access/security
▪️The biggest “right now” wins: document processing (invoices, orders, onboarding paperwork) and practical automation
▪️Why contact center AI is improving fast but still needs guardrails, human oversight, and smart UX
▪️The hard truth about hallucinations: why they’re tough to eliminate, and what actually helps
​
If you’re building voice AI, automation, agent-based systems, or just trying to cut through the hype—this one is jam packed!

ClueCon Weekly with Diego Gosmar