Mindee
by Mindee · Invoice & Document OCR
Developer-first OCR API that extracts structured data from invoices and documents.
- Works with
- QuickBooks, ERP / accounting systems (general), ERP / accounting systems (general)
- Deployment
- Cloud
- Company size
- SMB, Mid-market, Enterprise
- Pricing
- Tiered SaaS subscription plus pay-as-you-go credits (per page processed); free tier available
- Founded
- 2018
- Headquarters
- Paris, France
Overview
Mindee is a cloud-based, API-first document processing platform that uses OCR, computer vision, and machine learning to turn unstructured documents into structured JSON. It is built for developers and product teams who want to embed data extraction into their own applications and workflows rather than buy a packaged AP or capture suite. Mindee ships pre-trained models for common document types — invoices, receipts, bank statements, passports, ID cards, driver's licenses, resumes, proof of address, and barcodes — and also lets teams build custom models for proprietary document types. Output includes extracted field values, confidence scores, and bounding-box/polygon coordinates so results can be validated programmatically or routed for human review.
Beyond raw extraction, the platform provides document-handling primitives that address messy real-world inputs: Split breaks multi-page uploads into individual records, Classify routes incoming files by type, Crop isolates multiple documents scanned on a single page, and Chaining combines pre-processing and extraction steps in a single API call. Mindee positions itself as "training-free" for supported document types, meaning teams can call a pre-trained endpoint without assembling a labeled dataset first, while a continuous-learning capability lets models improve from feedback over time. The API accepts PDF, JPEG, and PNG inputs, supports multiple languages and alphabets, and offers asynchronous processing for large files.
Mindee is delivered as a REST API with official SDKs for Python, Node.js, Java, Ruby, and PHP, and connects to no-code automation platforms such as Zapier, Make, and n8n. For finance use cases, extracted invoice and receipt data is structured for export to ERP and accounting systems; Mindee does not publish certified, prebuilt ERP connectors but is integrated into downstream tools (for example pushing data to QuickBooks or Google Sheets) via the API and automation platforms. The company hosts in the EU and US and offers data-residency controls and "don't store my data" retention options on higher tiers.
Screenshots & demo
Demo video from the vendor's YouTube channel.
Features & capabilities
Data extraction
Core OCR and field-extraction engine.
- Pre-trained models for invoices, receipts, bank statements, passports, ID cards, driver's licenses, resumes, proof of address, and barcodes
- Extraction of header fields and line items from invoices and receipts
- Structured JSON output ready for downstream systems
- Per-field confidence scores for validation and review routing
- Bounding-box and polygon coordinates for each extracted value
- Support for multiple languages and alphabets
- Custom models for proprietary or unsupported document types
Document pre-processing
Primitives for handling messy, multi-document inputs.
- Split: separate multi-page uploads into distinct records
- Classify: route incoming documents by type
- Crop: isolate multiple documents scanned on a single page
- Chaining: combine pre-processing and extraction in one API call
- Input support for PDF, JPEG, and PNG formats
API and developer tooling
How teams integrate Mindee into their stack.
- REST API with JSON responses
- Official SDKs for Python, Node.js, Java, Ruby, and PHP
- Webhook support for asynchronous results
- Asynchronous processing for large files
- Live testing interface to try documents before integrating
- Developer documentation and API reference
Accuracy and continuous learning
Mechanisms for improving and verifying results.
- Training-free use of pre-trained models for supported document types
- Continuous-learning capability that improves models from feedback (RAG)
- Boosted-accuracy options on higher tiers
- Granular confidence scoring to flag low-confidence fields for review
No-code integrations
Connecting extracted data to business tools without code.
- Zapier integration to trigger extraction and route data to other apps
- Make integration for visual automation workflows
- n8n nodes for workflow automation
- Export of extracted data to tools such as QuickBooks, Google Sheets, Airtable, and Notion via automation platforms
Common use cases
- Accounts payable and invoice data capture
- Expense and receipt management automation
- Customer onboarding and KYC document verification
- Bank statement data extraction for lending and underwriting
- Resume/CV parsing for recruiting workflows
- Identity-document verification (passports, ID cards, driver's licenses)
- Embedding document extraction into SaaS products
Strengths & considerations
Strengths
- API-first, developer-oriented model designed to be embedded in custom applications
- Pre-trained, "training-free" models for common document types reduce setup versus template-based capture
- Transparent, public pay-as-you-go and tiered pricing with a free tier
- Per-field confidence scores plus bounding-box/polygon output for programmatic validation
- Pre-processing primitives (split, classify, crop, chaining) handle multi-document and multi-page inputs
- EU and US hosting with data-residency controls and 'don't store my data' options on higher tiers
Considerations
- No certified, prebuilt ERP connectors; ERP/accounting delivery is via API or no-code automation platforms
- Positioned as an extraction API rather than a full AP/capture suite with built-in workflow and approvals
- Reviewers note standard-plan support response times can be slow
- Password-protected or otherwise locked documents require pre-processing before extraction
- Advanced features (unlimited RAG, data localization, boosted accuracy) are gated to higher-priced tiers
ERP integrations
Extracted invoice/receipt data can be routed into QuickBooks via Zapier and similar automation platforms.
Structured JSON output is designed for export to ERP and accounting tools; integration is built via the REST API.
No-code delivery to downstream finance tools via Zapier, Make, and n8n.
Pricing
Free tier processes up to 250 pages/month with no credit card. Published paid tiers (annual billing) include Starter (~€44/mo, 500 credits, additional credits ~€0.05), Pro (~€179/mo, 2,500 credits, ~€0.04), and Business (~€584/mo, 10,000 credits, ~€0.035). Enterprise (250k+ credits/year) is custom-quoted. Credits are based on physical pages submitted regardless of document type. A 14-day free trial is available. Pricing figures observed at time of research and may change. Get an independent shortlist with pricing guidance below.
Technical & security
- Hosting
- Cloud (EU and US hosting options)
- Data residency
- European Union, United States
- Compliance
- SOC 2 Type II, GDPR
- Languages
- Multiple languages and alphabets supported
About the vendor
- Founded
- 2018
- Headquarters
- Paris, France
- Employees
- ~40
- Ownership
- Privately held; venture-backed (investors include GGV Capital, Alven, Serena Capital, Bpifrance)
- Notable customers
- Spendesk, Lucca, Indy, PayFit, Freedz
Alternatives to Mindee in Invoice & Document OCR
Mindee — frequently asked questions
How does Mindee integrate with ERP and accounting systems?
Mindee returns extracted document data as structured JSON via a REST API, which teams can push into ERP or accounting systems. It does not publish certified, prebuilt ERP connectors; instead, integration is built through the API or via no-code automation platforms such as Zapier, Make, and n8n, which can route data into tools like QuickBooks and Google Sheets.
What document types can Mindee process?
Mindee provides pre-trained models for invoices, receipts, bank statements, passports, ID cards, driver's licenses, resumes/CVs, proof of address, and barcodes, and supports custom models for proprietary document types. It accepts PDF, JPEG, and PNG inputs across multiple languages and alphabets.
Does Mindee require model training before use?
For its supported document types, Mindee is training-free — teams can call a pre-trained endpoint without assembling a labeled dataset. A continuous-learning capability can improve results from feedback over time, and custom models can be built for document types not covered by the pre-trained set.
How much does Mindee cost?
Mindee offers a free tier processing up to 250 pages per month with no credit card. Published paid plans (annual billing) start around €44/month (Starter, 500 credits) and scale through Pro and Business tiers, with additional credits priced per page. Enterprise plans for high volumes are custom-quoted, and a 14-day free trial is available. Pricing was observed at the time of research and may change.
What security and compliance standards does Mindee meet?
Mindee is SOC 2 Type II certified and GDPR-compliant. It offers EU and US hosting, and higher-tier plans add data-residency controls and 'don't store my data' retention options.
Evaluating Invoice & Document OCR?
Tell us your ERP and requirements and we'll send an independent shortlist — including Mindee and the best-fit alternatives — with honest pros and cons.