Baseplate (YC W23)

Backend-as-a-Service for LLM Apps
Baseplate is a backend optimized for LLM apps. Teams use our multimodal context database to build rich user experiences with LLMs. All from an interface as simple as a spreadsheet.
That means you don't have to maintain and manage separate databases for your vectors and regular data anymore!
Hybrid database for your regular data and your embeddings
Using Baseplate, a team can deploy a chat GPT app that responds with domain-specific information from documents, thumbnails, links, images, and more. (Who knows, maybe the next GPT model is multimodal too
Why would we need this?
In most applications, LLMs need to be connected to an ever changing set of data. It’s simple enough to parse and embed a few PDF’s. However, managing this data is tedious when you’re working with large, multimodal datasets that consistently need to be updated, re-indexed, and replaced.
Plus you’re likely managing multiple databases - one for vectors, and one for your other data. This gets painful really fast.
Key Features:
  1. 1.
    A flexible hybrid database. A single dataset in Baseplate can contain
    • Embeddings
    • Text
    • Code
    • Documents
    • Images
    • Links
  2. 2.
    Database management
    • Utilities for updating and managing vectors in bulk
    • Organize and segment your data intuitively
    • Work in the UI or programmatically via API
  3. 3.
    Smart Search
    • Choose an embedding model for your use case
    • Use keyword, semantic, or hybrid search based on your use case
    • Vector ranking based on more than just cosine similarity
      • Human Feedback, Last Updated, Last Accessed, etc.
  4. 4.
    • Build, test, and deploy LLMs via API endpoints
    • Unlimited queries
    • Built-in utilities for human feedback, logging, & caching


Last modified 9d ago