We're excited to introduce Spine, a tool we've been developing for the past month that aims to streamline the process of building and sharing AI-driven natural language interfaces like ChatGPT for various data formats.
With Spine, you can: * Upload numerous data types such as websites, PDFs, docs, PowerPoints, CSVs, audio files, YouTube videos, and more
* Navigate lots of data and get in-depth referenced results with our hybrid search
* Built-in Feedback: Craft accurate, personalized & adaptive experiences -- we will update the search models and eventually your LLM with the feedback
* Easy sharing: Seamlessly share your Spines so anyone can interact with your data (currently, you can share them on our website)
Spine is currently powered by a few open-source libraries (HF, llama-index, etc.) along with OpenAI. We have designed Spine so that it should be simple to make it LLM-agnostic; we should be able to swap out GPT-3 (the default behaviour right now) for Cohere or an open-source LLM like FLAN or Bloom. We also want to make it trivial to embed Spine in any form -- and are currently working on making an API available so users can embed us on their website or slack. We also have a few known issues we are navigating through: https://sharing.clickup.com/9003104561/l/h/6-900301350489-1/....
A bit about us: we're Akshay and Ashwin, two passionate tech folks with backgrounds in AI, ML, and startups. Akshay is an AI lead at Georgian, where he helps growth-stage companies build ML products and has previously worked at two YC companies. Ashwin is a senior engineer at Instacart with previous experience at Alexa & Aurora (AWS) and NVIDIA. We studied ML together at the University of Waterloo and Nanyang Technological University on an exchange program some years ago.
We created Spine in response to the growing demand for incorporating LLM-based chat interfaces into products and services in our companies. Our goal is to make this process as easy and accessible as possible.
We're still in beta and eager to listen and gather your feedback. To facilitate this, we've prepared three demos that can be accessed without signing up (search through guides for hiring or getting hired remotely, some material from the All-in crew, and a demo on education highlighting some multi-modality). We also have a couple of loom videos if you want to see Spine in Action on the landing page!
If you sign up (fingers crossed, everything runs smoothly), you can upload, create, and share as many Spines as you'd like for free!
Please share any feedback or suggestions you may have. We're all ears!