Rules to better ChatGPT and using the best API ever ⭐
summary
Demonstrates a working Rules GPT: company rules are preprocessed into embeddings as 1536‑dim vectors stored in a vector DB, looked up by cosine similarity, then injected into the system prompt and sent to the ChatGPT API; implemented as a Blazor app and delivering useful V2 email and scrum guidance. Estimates implementation as two developers two weeks for a custom build (or two days via Power Virtual Agents), notes GPT‑4 costs ~30× 3.5 and session/maintenance fees.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:06Speaker and talk overview
- 5:01ChatGPT news and extensions
- 16:57Prompt engineering techniques
- 22:42Rules with ChatGPT
- 27:17GPT API service
- 35:43Embeddings and vector databases
- 41:01Building the rules bot
- 49:31Power Virtual Agents vs solutions
- 58:44Future of AI and AutoGPT
- 1:02:17Summary
related talks
01h 09mChatGPT and OpenAI API solutions: 👑 My Favourite Implementations
01h 16mChatGPT - and How we Can Benefit from it
01h 01mHow ChatGPT Works
46mUsing AI in Software Design: How ChatGPT Can Help With Creating a Solution Arch.
01h 12mChatGPT and OpenAI API solutions: 👑 My Favourite Implementations
30mHow To Build Your Own Internal ChatGPT
38mIntroducing the New Colleague - ChatGPT
02h 00mChatGPT & Prompt Engineering