Babel

Summarize group chats and news into a personal daily report

Comment

Inspiration

  1. The world of Web3 is information overloaded. We have joined lots of group chats, but do not have time to read them all. So, we tend to miss out lots of important and first-hand information.
  2. The source of information or channels are complex, which makes it difficult to see all the updates from the influencers we follow on all platforms.
  3. The signal-to-noise ratio and information density are relatively low, resulting in low information intake efficiency, and it is time-consuming to find key info.
  4. We want to know the sources of information that others (especially bigwigs / KOLs) usually see.
  5. Get rid of the algorithm control of web2 giants, let people efficiently obtain the information they want, and avoid the fragmented information flood.

Combined with the NLP technical background of our team, we hope to have a product that aggregates our various public and private channels, summarizes high-value information, and sends us a personal summary daily report every day, so that we can read all the things in just 5 minutes.

How is it different from similar products?

What it does

Babel summarizes hundreds of chat messages into one sentence.

  1. Aggregate information from multi-channel sources, including Discord, twitter, clubhouse, Rss, etc. (constantly expanding)
  2. Summarize massive group chat messages into a streamlined real-time push
  3. Filter Twitter to only show tweets related to influencers talking about web3
  4. Summarize the common discussions, hotspots, and intelligence of each track into a personal daily report
  5. Trends, showing the hottest topics currently discussed
  6. Support cross-language summaries to clear reading barriers
  7. Personal Database (plan)
  8. Personal Daily NFT (plan)
  9. Android & iOS version (plan)

How we built it

  1. First, we get all the group chats a user has joined and all the messages in those chats in real time (for example, in Discord, we get all the servers, channels, and messages of the user). Discord/Twitter is currently supported, and more social media channels will be supported in the future.
  2. According to the chat heat/topic/number of messages and interval time and other dimensions, we will filter messages through the tuned NLP language model, summarize them in real time and push them to users. At the same time, identify and extract hot words and rank them according to the popularity, and present them in the "Trends" section.
  3. We will further classify and simplify the summary, and finally collect it into a customized daily newspaper and push it to users.

Challenges we ran into

  1. How to really improve SNR/information density from the technical/product strategy (how to accurately identify high-value information)
  2. How to make targeted summary according to users' circles/topics, and how to do quantification.

Accomplishments that we're proud of

The product demo can be used for trial at babel-web3.heyfriday.io/

We achieved

  1. AI language summary algorithm model was developed
  2. The MVP version of the product was officially launched, and the core functions have been open for experience
  3. Web3 senior seed users have given warm praises and show high expectations of our product

What's next for Babel

  1. Further refine our summary capability model and explore more summary presentation methods
  2. Upload daily to the public chain to provide users with the ability to cast NFT and open transactions. NFT holders can view the daily life of the foundry
  3. Open source or decentralize the model so that users can own their models
  4. Explore more possibilities of analyzing, forecasting and evaluating data
  5. Build a personal database to make Babel a treasure book of information aggregation for everyone
  6. Adapt to the mobile end

Built With

Share this project:

Updates