Launch HN: Baseplate (YC W23) – Back end-as-a-service for LLM apps

Baseplate is a unified backend-as-a-service platform for LLM applications that simplifies the infrastructure needed to build AI-powered products. It provides integrated data management, vector embeddings, prompt versioning, and deployment capabilities through a single API, eliminating the need for teams to build separate datasource integrations, vector databases, and storage solutions. The platform features a multimodal database, spreadsheet-style editing interface, hybrid search capabilities, and an App Builder for prompt iteration and context injection.

Key Points

• •Unified backend eliminates need to build separate stacks for datasource integrations, embedding jobs, vector databases, and storage—focus on core product instead
• •Multimodal database stores text, embeddings, metadata, and images in one place with spreadsheet-style editing interface for easy vector/metadata management
• •Hybrid search combines semantic and keyword-based search using tunable open-source models (Instructor & SPLADE) for different dataset types
• •App Builder enables iterative prompt engineering with input variables and context variables that dynamically pull from datasets at query time
• •Single API endpoint deployment pulls from one or multiple datasets, injects context into prompts, and returns search results for custom UX building
• •Built-in human feedback and logging utilities for monitoring and improving deployed endpoints
• •Team workspace enables collaborative iteration and prompt management across team members
• •Datasource sync via API or UI keeps datasets fresh from Google Drive, Notion, and other sources
• •Pricing: $35/month per user (Pro) or $500/team (Team plan) with self-serve signup
• •Future roadmap includes piping images directly to multimodal models like GPT-4 for enhanced capabilities

Workflow Diagram

Concepts