OmniAI (YC W24)

Omni is building an ETL for unstructured data. Tech stack is TypeScript | Postgres | Docker | React | and about 5 different LLMs. The main things we spend our time on: 1. Wrangling LLMs into providing predictable outputs 2. Running data transforms at scale 3. Processing multimodal data (audio, documents, images, etc.) All of these problems are hard, especially in conjunction with each other. If you’ve had any experience with structured LLM output, we’d love to chat.