Introduction Building production-ready AI agents requires more than just a prompt and an LLM. You need infrastructure for state management, tool execution, and secure deployment. I recently built https://ai-secure.dev, a SaaS for automated security…

Learnings Try to have complex flows solely solved by the agent. And then write deterministic code. speed. Introduction Building production-ready AI agents requires more than just a prompt and an LLM. You need infrastructure for state management, tool…

Recently, AWS released Claude v3. It comes with the Haiku and Sonnet flavors. Both are big improvements over the previous version. We recently updated our arcBot to use Claude v3 Sonnet and are very impressed with the results. Not only are the…

Using the AWS Bedrock API for MVPs is incredibly enjoyable! I recently wrote an article on how you can make the LLM Claude respond in JSON. You can check it out here. While it's a lot of fun, testing your LLM settings and prompts can become tiresome…

Validating the response from your Language Learning Model (LLM) is a critical step in the development process. It ensures that the response is in the correct format and contains the expected data. Manual evaluation can quickly become tiresome…

Working with the AWS Bedrock API is an exhilarating experience! I came across an interesting business case where I needed to develop an AI MVP. The MVP generates JSON data based on a prompt and utilizes the anthropic.claude-v2:1 model in AWS Bedrock…