Building Multi-Tenant SaaS Architectures

Chapter 16. GenAI and Multi-Tenancy

The whole world of software development is asking itself how and where they can introduce generative AI (GenAI) into their offerings. GenAI has opened an entire new front of opportunity that has teams evaluating how and where they can introduce GenAI constructs into their architecture. If we look at the potential of GenAI, you can imagine that it could have profound impacts across applications from a wide range of domains and use cases. For our purposes, though, I wanted to identify areas where SaaS providers could mix GenAI constructs with multi-tenancy to deliver new experiences that could differentiate their offerings. With that in mind, I’ve focused this chapter on outlining specific GenAI architecture strategies that enable SaaS providers to introduce tenant contextual capabilities into their GenAI model. This added context would allow SaaS providers to have a single, shared multi-tenant GenAI footprint that applies tenant context to inference, yielding responses that are tailored to the needs of individual tenants. This also opens an entirely new landscape of multi-tenant considerations. Isolation, noisy neighbor, cost, and pricing are new territory that will require new approaches to applying these principles in a multi-tenant context.

To get started, though, we’ll need to establish a foundation, reviewing some of the core concepts that come with building a multi-tenant solution that includes GenAI-enabled capabilities. The goal here ...