Superpowered AI - Knowledge Base as a Service for LLM applications

Hey everyone! We are Zach McCormick, Dillon Martin, Justin Clark, and Nick McCormick from Superpowered AI.

TLDR

Superpowered AI is an end-to-end knowledge retrieval solution designed specifically for LLM applications (i.e. retrieval-augmented generation). We turn complex infrastructure into a few simple API calls, making it easy to retrieve the right information and deliver highly relevant responses to your users.

🦉 Who is it for?

Whether you’re using LLMs to improve your customer experience, increase efficiency within your company, or launch an LLM-based application for end-users, Superpowered AI can provide your app with external knowledge and long-term memory without requiring you to be an AI expert – all at a low cost. How low? We’ve set our pricing at half the cost of popular alternatives like OpenAI embeddings + Pinecone. Please note, our product is completely free for now, so feel free to query your heart out!

💡 How does it work?

Simply upload files (including audio) to a Knowledge Base, and then you can begin querying! We handle the text splitting, embeddings calculation, vector database uploading, etc. for you. When you query your Superpowered AI Knowledge Bases, we return a list of the most relevant snippets of text from your Knowledge Bases, along with an LLM-generated summary of the results (with citations). This output can then be added to the prompt of your LLM-based application to dramatically improve your LLM responses. Accepted file types: .pdf, .docx, .txt, .md, .mp3, .m4a, .wav

You can check this out in our web app right now!

Or, you can use our SDK 🤓 :

Step 1: Install Superpowered SDK

Step 2: Create a Knowledge Base

Step 3: Query from Knowledge Base

👊 Why we built Superpowered AI

LLMs have emerged as a groundbreaking new technology and we are thrilled to be part of this revolution! But when LLMs are used on their own they have major shortcomings. For real-world use cases, we could see the importance of integrating additional functionality into them to overcome these issues.

1. Access to external knowledge. LLMs lack reliable knowledge and have none of your files to work with, but providing them access to knowledge will change your LLM app from a “neat toy” to a powerful tool for real-world use.
2. Long-term memory. Want to build a chatbot that remembers users’ preferences over long periods of time? You’ll need to store conversations and retrieve them later on.

🙏 Our Ask

1. Check out Superpowered AI and join our Discord!
2. Talk to us about your use case. If you’re building any kind of LLM application, there’s a very good chance we can help!

🚀 Sign up & chat with us:

Create your account and/or book a demo at superpowered.ai, or email me at dillon@superpowered.ai