One Prompt is NOT enough: Using AutoGen to code a Multi-Agent Postgres AI Tool

Video Description

There's only so much you can do with a single prompt. There comes a point in time where you can only do so much with prompt engineering before you hit a wall. Thankfully the code bros at Microsoft have been cooking up AUTOGEN. A simple way to build prompts that work together to solve a problem. AKA a multi-agent framework for problem solving. In this video we enhance our AI charged Postgres Data Analytics agent backed by GPT-4 and we make it MULTI-AGENT. By splitting up our BI analytics tool into separate agents we can assign individual roles as if our AI was a small working software data analytics company. We build a data analytics agent, a Sr Data Analytics agent, and a Product Manager Agent. Each agent has a specific role and we can assign them special functions that only they can run. Of course, we utilize our favorite AI pair programming assistant AIDER to generate a first pass of our code in no time with the help of a couple prompt engineering techniques. We build in python and use poetry as our dependency manager. Our goal is to move closer to the future of AI engineering and build a fully functional AI powered data analytics tool with ZERO code. Agentic software is likely the future, so let's stay on the edge of AI engineering and build a multi-agent data analytics tool with AutoGen. 👍 THE CODEBASE https://github.com/disler/multi-agent-postgres-data-analytics/tree/v2-using-autogen-to-build-our-multi-agent-tool ✅ Watch Part One - building a Postgres Data Analytics Agent From Scratch https://youtu.be/jmDMusirPKA 🤖💻 AI Engineering Resources Microsoft's Autogen: https://microsoft.github.io/autogen/ Autogen group chat example: https://github.com/microsoft/autogen/blob/main/notebook/agentchat_groupchat_research.ipynb AIDER: https://aider.chat/ Free Postgres Hosting With Neon: https://neon.tech/ 🤖 ZERO Touch coding with AIDER? https://youtu.be/MPYFPvxfGZs 📘 Chapters 00:00 One MAJOR Problem 00:45 Okay so what is AutoGen? 02:00 Make our app Multi-Agent with AutoGen 05:00 Defining our agents 06:10 Let AI Code for you with AIDER 08:30 Half way there let's clean it up 14:04 Our Multi-Agent Analytics App is Born 18:40 Is AutoGen The Future of AI Engineering? 20:30 AutoGen Pros 21:54 AutoGen Cons 23:10 My take and what's next 🐛 tags #dataanalytics #agentic #promptengineering