In time of despair, I like to remember what Albert Camus said about being a rebel What is a rebel? A man who says no, but whose refusal does not imply a renunciation. He is also a man who says yes, from the moment he makes his first gesture of rebellion.
You might be interested in Graphiti: https://github.com/getzep/graphiti.
With a self-hosted Graphiti MCP, you can connect ChatGPT or Claude to build a knowledge graph from all your data. You can then query and update the graph directly through conversation & by ingesting data and visualize the graph using tools like the Neo4j Explorer.Don’t know if that could fit your use case but that could be a fun way!
As someone who has loved sagrada familia since I went in, I think the experience of Sainte Chapelle is my second favorite (go first thing when it opens to have it to yourself) and is more underrated than sagrada familia^
Related/unrelated, part of my joy in the sagrada familia is that being a tourist feels essentially the same as being a pilgrim. If you get a chance to visit parc guell, you aren't exactly experiencing it as a park, but as a tour through the different ideas in the park. (Compare this with an unguided stroll through Central Park, where you and all of the other visitors are likely experiencing it as a park (the way it was intended)
^ I think! In my experience it's occasionally overlooked in a short trip to Paris, whereas if you're going to just see one Gaudi, make it the cathedral
Great idea! How does it compare to Graphiti / Zep? They use real time knowledge-graphs to tackle the problem you are trying to solve https://github.com/getzep/graphiti
Interesting read, I searched if another startup got this market right, it seems truelayer did using open banking for online payment ($700m valuation) https://truelayer.com/
The issue isn't search or LLMs, it's the data sources. I believe we can create something better than the traditional web for sharing and accessing information in the LLM era. What do you think?
Hmm i think the issue with any large data set is indexing + querying. At https://usefind.ai/ this is a fundamental problem we've been trying to tackle. What kind of structure do you think would work?
Hmm i'll check out the pagerank stuff, IMO RAG isn't super great. I think RAG overall has been oversold since embeddings KNN hasn't proven to be super accurate.
I completely agree with you. Establishing clear guidelines on how data is used and who can access it would be essential. This concept has value beyond just sales and marketing. For instance, if someone is seeking funding today, where can they find a comprehensive list of potential funders, determine their eligibility, understand the application process, and know what documents are required?
Of course, spam could be an issue. To maintain a trustworthy and reliable environment, a robust reputation system—similar to Google's PageRank could evaluate individuals and their contributions. It would replicate the way we naturally seek advice from trusted individuals but on a global, digital scale.
I was considering creating a data cooperative - democratically owned and governed by its members. To ensure sustainability, revenue could come from AI companies and other stakeholders paying for data access, much like how news publishers are compensated today. What do you think?
