Episode 35 · Season 1
Agentic AI Boundaries: Multi-Tenancy Done Right
Multi-tenancy in agentic AI is not the same thing as multi-user, and getting it wrong can create serious security, compliance, and data boundary risks. In this episode, Greg Banks and Gavin Franklin break down what actually changes when an AI agent serves multiple users vs multiple tenants.
What you’ll learn
- How to distinguish individual, multi-user, and multi-tenant agent architectures (and why multi-tenant doesn’t always imply multi-user).
- Where cross-tenant leaks happen: memory, embeddings + vector search, and logs/traces—and what “tenant-scoped” really means.
- Why shared OAuth/API tokens and “UI-only filtering” are dangerous—and how MCP-style services help reduce risk.
Who it’s for
Business leaders and technical teams designing, governing, or securing agentic AI systems with real enterprise data.
Guest + credibility
Gavin Franklin joins host Greg Banks to explain the architecture and security controls that matter most when agents operate across users and tenants.
CTA
Have questions or want a follow-up topic? Email CirriusTalk@Cirriussolutions.com. (And follow the show for more on agentic AI.)
For questions or feedback, please contact us at CirriusTalk@CirriusSolutions.com
Recommended Links
Episode Artifact:
Gavin Franklin on LinkedIn
Greg Banks on LinkedIn
Jason Fowler Music
Cirrius Solutions
We cover
- Understanding Multi-Tenancy vs Multi-User
- Defining Key Concepts in AI Agents
- Memory Management in Multi-Tenant Environments
- Document Handling and Security Concerns
- Logging and Tracing in AI Systems
- API Security and Multi-Tenancy
Chapters
- 0:00Understanding Multi-Tenancy vs Multi-User
- 2:21Defining Key Concepts in AI Agents
- 5:09Memory Management in Multi-Tenant Environments
- 7:51Document Handling and Security Concerns
- 10:31Logging and Tracing in AI Systems
- 13:08API Security and Multi-Tenancy
- 15:56Tenant Offboarding and Data Management
- 18:46Navigating GDPR and Privacy Laws
- 21:33Final Thoughts and Key Takeaways
Resources and downloads
The artifacts Greg’s team built for this episode, free to download.
Agentic AI Boundaries: Multi-Tenancy Done Right
This document provides a side-by-side comparison of multi-tier versus multi-tenancy architecture; failing to consider the difference can cause data leaking.
Related episodes
Agentic AIPrompt Injection: 4 Practical Defenses for Your AI Agents
Agentic AIAgentic AI – Local vs Cloud-Hosted Agents
Agentic AIBuild Your First Custom AI Agent: A Step-by-Step Walkthrough
The Military MinutePurpose After Service Jon Macaskill on Military Transition & MindfulnessJon Macaskill
Salesforce at ScaleMinimum Viable Product for Salesforce: A Phased Implementation PlaybookTalk to us about AI agents
Keep going: See how we build AI agents for real businesses, Talk to us about a custom AI agent, More from the Cirrius blog