Dear UIUC.chat users,

You’re receiving this email since you’ve signed up for UIUC.chat. Here’s our first ever newsletter! We will only email you about product updates. I hope this is ‘high alpha’ insight into the best uses of LLMs that we’re pushing into production.

We’re a small team of academics: 2 full-time researchers at NCSA, 4 grad students, and a few undergrads. Everything we build is free!

New Features

Launching GPT-4o & Llama3

We now support GPT-4o; just “bring your own” OpenAI API key. It’s super fast and can access all of your knowledge via the documents you upload.

We’re hosting Llama3 for free! Register here for access when it launches. No more pesky OpenAI bills! 🎉

Web Crawling

Our web crawling + scraping is getting seriously powerful:

• It’s crazy fast (upwards of 2GB/s of crawling speed!)
• It does full JavaScript rendering, ensuring no content is missed.
• We grab full PDF files while crawling.
• The crawler now finds every possible page on a website using the open-source crawlee.dev.

Read the docs →

Access it on your Materials page: uiuc.chat/YOUR_PROJECT/materials.

Here’s what to look for. Just enter any URL to crawl that site. It runs in the background. Feel free to start multiple crawls from different parts on the same website to get full coverage.

Document Groups

Document Groups help you find what you need faster. You can:

• Organize documents into clear, manageable categories.
• Enable/disable Document Groups to control what your chatbot users see.
• Chat with subsets of your data using filters (coming soon).

Try it out and start navigating huge projects with ease and control 🙌.
Access it on your Materials page: uiuc.chat/YOUR_PROJECT/materials.

Fresh Docs Just Dropped!

Dive into our updated, super-detailed documentation to get the most out of UIUC.chat.

Explore the new docs here →

Want the nitty-gritty? Check out our System Architecture, or current hosting costs, or read about our retrieval systems (RAG).

More Features Coming Soon

1. Custom tool use by the LLMs: Drag-n-drop workflow builder interface thanks to n8n.io. Way easier than OpenAI’s tools.
   1. Getting data into your chat: Imagine a real-time API connection to your email, Jira, or Notion. The LLM can query these as needed using function calling.
   2. Taking action from chat: After analyzing data, authorize the LLM to send emails or update Jira tickets.
2. LLM Guided Retrieval: We’re developing features to help you find relevant documents in a sea of millions.We've just submitted a grant proposal to Amazon’s Illinois Center to fund development, but we’ve already started building it.

Watch or Share our Product Demo

If you have any feedback, please email Kastan at [email protected]. I’m happy to help you get the most out of UIUC.chat. If you’re internal to UIUC, email me and I’ll get you free OpenAI credits ;)

Thank you for your continued support, and we look forward to serving you with even more exciting updates in the future!

Best,
Kastan & UIUC.chat Team