⚖️ In my conversation with customers, I don't get tired in highlighting how important it is to decompose

⚖️ In my conversation with customers, I don’t get tired in highlighting how important it is to decompose application’s needs into the different use cases and choose the best-fitting foundation model per use case, balancing functional fit, cost and frugal resource utilisation — trade-offs to weigh per workload, not a single optimum.

🔐 This is all good, but obviously increases the complexity of the overall solution slightly and leaves us with the need to define guardrails across different foundation models to secure the model inputs and outputs of our applications — one layer of application security, not a replacement for the rest. Here Amazon Bedrock guardrails come to the rescue. The allow to define Guardrails spanning across different foundation models (see the Amazon Bedrock Guardrails documentation). You can have Bedrock apply them directly when invoking a model, or call the guardrail evaluation via its own API from your application, wherever the content originates. In my re:invent recap(https://www.linkedin.com/pulse/what-happens-las-vegas-nah-lets-have-look-all-things-christoph-vltke/?trackingId=FZeh33BNRR%2BZ7yY3nbu9vQ%3D%3D), I highlighted how new functionality in Amazon Bedrock Guardrails makes this approach even more powerful. But a Markus points out - it also got now so much more affordable. So no excuses 😉

🤿 Like always - ping me if you want a deep dive!

• AI
• AWS
• Security