Prompt Shepard: AI Agents with Determinstic Outputs

​This will not be a slideshow presentation, we're gonna do some real live coding, so please some with VSCode/Cursor ready to go!

​What the coding session will entail:

1. ​We'll build an Agent that composes multiple LLM calls to get higher accuracy (we may use nextjs or python based on the audience polls)

2. ​We'll cover some popular hallucination reduction techniques

3. ​We'll talk about how to hit reliability

​We may go an extra 30 minutes for any Q&A.

​Code and Recording will be uploaded to public viewing afterwars.

​See the full agenda here: https://docs.google.com/document/d/1cF-hrFsBTvlQyRd5Sc7PpX8ndSy3-r9KyXTPMbCG0gw/edit?usp=sharing