Nanonets Document Intelligence and AI Agents

Nanonets is a San Francisco-based AI company backed by Y Combinator, Accel and Elevation Capital. It reports more than 10,000 customers, including 34% of Fortune 500 companies, and over 1 billion documents processed. It targets the manual handling of invoices, purchase orders, claims and contracts in finance and operations teams.

The Document Intelligence product extracts structured data from PDFs, images, Word, Excel and email without templates, classifies and splits multi-document files, and keeps tables and line items intact. Each field carries a confidence score, and low-confidence fields go to human review. Nanonets builds its own OCR models, including the open Nanonets-OCR2 family and OCR-3, and adds agents that apply business rules and write results into systems of record such as SAP. Output is available as JSON, Markdown or CSV via API.

• Template-free extraction: pulls fields, tables and line items from PDFs, images, Office files and email.
• Classification and splitting: separates multi-document batches of up to hundreds of pages.
• Confidence scores and human review: routes low-confidence fields to reviewers.
• Own OCR models: Nanonets-OCR2 and OCR-3 convert documents to markdown and answer questions about them.
• AI agents: apply business rules and write data into SAP and other systems of record.
• API and VPC deployment: parse, extract and split endpoints with private-cloud option.

Vietnamese distributors can automate supplier invoice capture and three-way matching before posting to SAP. Logistics companies can extract data from bills of lading, packing lists and customs forms. Software firms can embed the extraction API into accounting or lending apps for Southeast Asian customers.

The product page states 99%+ field accuracy on production document types and support for more than 100 languages, including mixed-script and handwritten documents. Confidence scores and validation rules send only uncertain fields to people, cutting review work. Deployment in a customer VPC supports strict data residency requirements.

Document extraction is moving from template OCR to vision-language models and agents that complete the downstream workflow, such as matching invoices to purchase orders. Companies across Southeast Asia that run accounts payable, order management and logistics in several languages need template-free, multilingual extraction delivered as an API.

Nanonets lists its address at 2261 Market Street, San Francisco, and investors including Y Combinator, Accel, Elevation Capital, SV Angel and Avalon Ventures. In March 2024 it announced a 29 million USD Series B led by Accel India, bringing total funding to 42 million USD. It states SOC 2, HIPAA, GDPR and ISO compliance and publishes open OCR models on Hugging Face.