At Info Support, we use agents a lot. This comes with a sizable credit card bill and introduces challenges around rate limits, availability, billing, and more. We’ve adopted an interesting approach to this by building an AI platform strategy that combines cloud computing across multiple providers with locally hosted LLMs in our data center. In this session, we’ll explore the challenges we faced when adopting AI tools at our organization, why we decided to combine cloud computing with locally hosted models, and how we’re building our platform to support this. We’ll not only explore technical and architectural patterns, but also examine how we combine platform architecture with our way of working to maximize the positive impact on our work.