
- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
‎May 20, 2024 03:21 PM
Hello,
AirTable introduce AI Feature few months ago but I don"t see Vector Search or Semantic Search.
I found this project Ask-Airy that provide a very interesting design a year ago.
- It create an Embeding for each table's row
- It create a HyDE of Row based on user query to build an Embeding
- It search the results using Cosine Similarity (I'm struggle here)
I like the way it is done but I lack of explanations. I think AirTable should provide this kind of feature natively.
As a GPT developers I'm looking for RAG applications.
- Yes I could use a SaaS database like Pinecore
- Yes I could use Azure / OpenAI Search APIs
- But what about AirTable ?
I have a lot of interesting Data into AirTable, I'd love to simply query AirTable with Semantic or Vector search then get top X results and fill back my LLM tools.
Any ideas ?
- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
‎Jun 15, 2024 12:10 AM
GPTs (Generative Pre-trained Transformers) are revolutionizing semantic search by understanding natural language queries and retrieving relevant information from vast datasets. xfinity billing customer support With their ability to comprehend context and semantics, GPTs enhance search accuracy, enabling users to find precise answers to complex queries. By leveraging advanced neural network architectures, semantic search powered by GPTs offers a more intuitive and efficient way to access knowledge across diverse domains and applications.
- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
‎Oct 24, 2024 12:29 PM
Hello @Jean-Philippe_E
Did you get any relevant help on your question? I am very interested by this topic as well.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
‎Oct 24, 2024 12:32 PM
I store embedding into a cell of each row, then perform cosine similarties along all rows when I search for something
