In this episode, Dr. Sanjiva Weerawarana and Asanka Abeysinghe demystify what “agents” really are and why architects should care. They walk through core concepts and terminology—agents, agent loops, prompts, context, memory, RAG, tools, MCP, and skills—and discuss how agents observe, act, and evaluate. The conversation compares agents to traditional systems, explores where agents fit in modern architectures (including solo agents, agent-to-agent patterns, and multi-agent setups), and looks at orchestration challenges. They also cover risks, guardrails, and governance, and close with practical takeaways on what architects should start doing now, with a hook into the next episode on the agentic enterprise.

🔥 Key topics covered:

• What are ‘agents’

• Terms: Agent, agent loop (e.g. chain of thought), prompt (system prompt, role, user prompt), context, context graph, memory, RAG, tools, MCP, skills, observe & test → evaluate, interfaces (or how to talk to the agent), solo agents, A2A, multi-agents, agent orchestration

• Agents vs traditional systems 

• Where agents fit in

• Risks, guardrails and governance 

• What architects should do now 

• Hook to the next session, ‘Agentic Enterprise.’

🎙️Hosts

Sanjiva Weerawarana

LinkedIn: https://www.linkedin.com/in/sanjivaweerawarana/

Twitter: @sanjiva

Medium: https://medium.com/@sanjiva.weerawarana

Asanka Abeysinghe

LinkedIn:https://www.linkedin.com/in/asankaabeysinghe

Twitter: @asankama

Blog: https://blog.architect2architect.com/

Medium: https://asankama.medium.com/

#wso2 #genai #microservices