Mixture of Predictive Agents (MoPA) - The Wisdom of Many AI Agents Architecture

Mixture of Predictive Agents (MoPA) - The Wisdom of Many AI Agents Architecture

👊 Become a member and get access to GitHub, Code and Discord:

🤖 Great AI Engineer Course:

🔥 Open GitHub Repos:

📧 Join the newsletter:

🌐 My website:

Today we take a look at my Mixture of Predictive Agents (MoPA) architecture. This will use the wisdom of many theory to try to predict a outcome by taking the avg of all outputs from many llm models.

00:00 MoPA Architecture
02:37 Hubspot
03:41 MoPA Python Code
10:47 Testing the MoPA System
13:41 Final Thoughts

[AlenaHampsteadsmith]

Predictive deep learning technology is the BOMB 💣💣💣 make sure to add Lemon AI into your mixture, if you need to optimize ad campaigns and bring your marketing game to the next level

[Nifty-Stuff]

This architecture is BRILLIANT and so close to my dream system/process! For the last year I've wondered: "Why hasn't anybody developed a system/app that takes API's from the top LLMs, and created agents for each... and then have these agents all work together to brainstorm, debate, review, and solve problems, and then present me with the best solution / answer that they (mostly) agree on?!?!?"

I often get 4 different answers from 4 LLMs, so why not have them all setup as agents "in one room" working together to come up with the "best" solution. I can't find anybody that's tried this... why not? Wouldn't having the "top minds" (LLMs) working together produce better results?

I have to imagine some very creative coders could create the logic, limits, instructions, roles, etc., to guide how the LLMs interact, balancing enough "discussion" to get to a good answer. Your architecture is so close to this!

[paelnever]

In real world bitcoin prices depend on thousands of different factors like energy price or asic hardware price, it would be much more clever to analyze those factors than simply the past prices.

[klammer]

Great work as always Kris! Keep em coming!🥳🤩🦾

[jasonedward]

I love the idea that you could aggregate predictions and then back test on past data to course correct. Thinking what else this could apply to besides trading

[dawmro]

I am curious what would happen if you don't tell the model to be slightly negative, would the price prediction be better?

[watchdog]

How you gonna predict based on 30 day historical prices? You need tons of measurements, like social media presence, price since beginning of time, wallets created, technology interest, fear & greed, etc etc etc.

[ryanjames]

here we go, to the moon !!! great work thanks for sharing,

[ewasteredux]

Hey Kris! I was looking for the source for this (MoPA) and am having issues locating it. Am I just overlooking it or has it not been uploaded yet?

[fnice]

Did you know? that chatGPT can do what other models don't offer yet? creating full project and then zip's the files and folders, largest project chatgpt has created for me, was about 6mb zipped, unzipped little over 15mb project files and folders. it's all about the prompt's. I'm sure much larger projects could be created with the correct prompt's.