Features
Agentic Prompt Studio for automated schema generation, prompt crafting, and accuracy validation
Prompt Engineering Interface for manual extraction prompt design with versioning
Human-in-the-Loop verification for accuracy control
Single Pass & Summarized Extraction modes (reduces token usage up to 7x)
LLMWhisperer OCR engine with four processing modes: Native Text, Low Cost, High Quality, High Quality with Form Elements
Multi-agent table extraction (6 agents + codegen step) for complex layouts
Pre-built API endpoints: Table Extraction, PDF Splitter, Invoice, Bank Statement, Purchase Order, Bill of Lading
MCP Server for integration with existing stacks (e.g., n8n)
LLMChallenge: two-LLM consensus to reduce hallucinations
Bring your own keys for LLMs, vector databases, and embedding models
Deployment options: managed cloud, on-premise, open-source
Document agnostic – works without prior training or templates
Supports PDFs, images, MS Office, LibreOffice, scanned documents, handwritten forms
Versioning built-in for extraction prompts
Multi-LLM comparison for cost and response quality
=SHEETAI() text generation in cells
=SHEETAI_LIST() list generation
=SHEETAI_EXTRACT() data extraction from text
=SHEETAI_CLASSIFY() text classification
=SHEETAI_FILL() auto-fill columns with AI
=SHEETAI_SUMMARIZE() text summarization
Supports GPT-4, Claude, Gemini models
Bring your own API key (OpenAI, Claude, Replicate)
Cache feature to reduce API costs
Free tier with limited calls
Paid plans from $20/month
