✦
Summary
Trellis is an AI platform that turns unstructured data into structured, queryable tables, described as an LLM-powered Snowflake for unstructured data. It converts PDFs, emails, call transcripts, contracts and images into SQL-compliant tables with schemas defined in plain natural language, then lets you run SQL queries directly against those sources.
Key Features
Converts unstructured data into SQL-compliant tables
Schemas defined in natural language
Run SQL queries directly on documents, emails and contracts
Multimodal models that process text, images, PDFs and HTML
OCR and AI vision for scanned documents and tables
Syncs with 200+ data sources
Flexible API and no-code interfaces
Live demo sandbox with no sign-in required
About Trellis
It uses multimodal models with OCR, table transformation and context-aware chunking to extract accurate, auditable data, syncs with more than 200 data sources, and offers both a flexible API and no-code interfaces, with a live demo sandbox that needs no sign-in.
It suits data teams and analysts who need to query documents and messages as if they were databases, and it was founded as a Y Combinator-backed Stanford AI Lab spinout.