Five Thoughts on AI Agents

There is a lot of excitement around AI Agents.  Here are five questions we're asking ourselves about how AI Agents will be used and managed. 

SHOW: 918

SHOW TRANSCRIPT: The Cloudcast #918 Transcript

SHOW VIDEO: https://youtube.com/@TheCloudcastNET 

CLOUD NEWS OF THE WEEK: http://bit.ly/cloudcast-cnotw

CHECK OUT OUR NEW PODCAST: "CLOUDCAST BASICS"

SHOW SPONSORS:

• Cut Enterprise IT Support Costs by 30-50% with US Cloud

SHOW NOTES:

• Anthropic warns that fully AI employees are a year away
• AuthZ Agents Chaos
• Why we need to get AuthZ right
• AI Agent Gateway and AI Agent Mesh
• Auth in the age of AI Agent (Cloudcast Eps. 885)
• Trust will make or break AI Agents 
• Introducing the Agent2Agent Protocol (A2A - Google)
• Introducing the Model Context Protocol (MCP - Anthroopic)
• Top 9 AI Agent Frameworks

HOW WILL WE VIEW AN AI AGENT IN THE CONTEXT OF HUMANS OR “USERS”

• [Who] How is an AI agent represented in the context of a person, a group, an organization? 
• [What] How will we audit Agent created data or content? Will we need a way to label Agent-created information?
• [Where] Will we have agents that are focused on non-human-interactions (e.g. optimizations) vs. human-interactions (e.g. empathy)?
• [When] Are today’s auth systems able to manage AI agents, or will they need to be re-engineered to deal with different scale, granularity of actions?
• [Why] Why will we choose an Agent (as the implementation) over another form or working on a problem?
• [How] Are today’s auth systems able to manage AI agents, or will they need to be re-engineered to deal with different scale, granularity of actions?

FEEDBACK?

• Email: show at the cloudcast dot net
• Twitter/X: @cloudcastpod
• BlueSky: @cloudcastpod.bsky.social
• Instagram: @cloudcastpod
• TikTok: @cloudcastpod