What Document Formats Does Landing AI ADE Support?

Landing AI ADE supports PDF, PNG, JPG, TIFF, BMP, HEIC, DOC, DOCX, PPT, PPTX, ODT, CSV, and XLSX. Text documents and presentations are converted to PDF before parsing. Spreadsheets are parsed sheet-by-sheet with embedded image extraction.

How Does Landing AI Handle Multi-Page Table Extraction?

ADE uses cross-page table reconstruction to automatically merge tables that span page breaks into a single structured array. This eliminates the need for manual post-processing or custom merging logic in downstream pipelines.

What Is The DPT-2 Parsing Model In Landing AI?

DPT-2 (Document Pre-Trained Transformer) is Landing AI's proprietary parsing model that identifies document layouts, chunks content, and generates captions for figures. Users can pin specific model snapshots for consistent results or use the latest version for automatic improvements.

Can Landing AI Parse Handwritten And Multilingual Documents?

Yes, ADE can parse handwritten text across multiple languages. The vision-first architecture reads document content visually, enabling extraction from handwritten notes, multilingual forms, and mixed-language documents without language-specific configuration.

How Does Confidence Scoring Work In Landing AI ADE?

Each parsed chunk and table cell receives a confidence score from 0.0 to 1.0. Scores below 0.95 are flagged as low confidence with exact character positions highlighted. This allows teams to route uncertain content for human review programmatically via the API.

Landing AI

Name: Landing AI
Availability: InStock
Author: LandingAI

by LandingAI • Mountain View, CA, USA • Founded 2017

Vision-First Agentic Document Extraction for Production-Grade Enterprise AI

No reviews yet

10 4

document-ai computer-vision document-extraction enterprise-ai api

Visit Landing AI

Follow:

Trust Score

Based on ratings & reviews

4.6 /5

40 reviews

Views

Clicks

Pricing

From $250/mo

What is Landing AI?

Landing AI is an API-first document intelligence platform founded by Andrew Ng, co-founder of Coursera and founding lead of Google Brain. Its flagship product, Agentic Document Extraction (ADE), converts complex documents like PDFs, spreadsheets, and presentations into structured, machine-readable data with full audit traceability.

ADE uses proprietary vision models to reliably parse dense layouts, multi-page tables, and multilingual content at thousands of pages per minute. The platform supports modular REST APIs, Python and TypeScript SDKs, and offers confidence scoring, visual grounding, and zero data retention for regulated industries including finance, healthcare, insurance, and legal.

Landing AI Demo Video

Key Features 8

Agentic Document Extraction API for Accurate Structured Data From Complex Documents

DPT-2 Vision Model for Tables, Signatures, Checkboxes, and QR Codes

LandingLens No-Code Platform for Building Custom Computer Vision Models Fast

Visual Prompting to Create and Deploy Computer Vision Projects in Minutes

Schema-Driven Data Extraction With Confidence Scores and Audit-Ready Traceability

Data-Centric AI Approach With Automated Mislabel Detection and Collaborative Labeling

Flexible Cloud, On-Premises, and Edge Device Deployment for Enterprise Environments

SOC 2 Type II, GDPR, and HIPAA Compliant With Zero Data Retention Option

Who Is Landing AI For

1 Financial Services Analysts

2 Compliance Officers

3 Document Automation Engineers

4 Healthcare Data Teams

5 Legal Operations Professionals

6 AI/ML Developers

Pros & Cons

Pros 4 benefits

Vision-First Parsing Architecture
Audit-Ready Visual Grounding
Founder-Led AI Expertise
Flexible Credit-Based Pricing

Cons 3 limitations

No Native UI Workflow Builder
Enterprise-Only VPC Deployment
Limited Non-Document AI Scope

Frequently Asked Questions

5 questions

Trusted Reviews

Verified Platforms

4.6 G2 (4)

What's New

weekly

New APIs: ADE Classify and ADE Section

Two new APIs in public preview: ADE Classify assigns a class to each page for downstream routing, and ADE Section generates a hierarchical table of contents from parsed documents.

Apr 22

New Parsing Model Snapshot: dpt-2-20260410 dpt-2-20260410

Improved cell parsing in forms and tables with better column alignment in complex tables. Text positioned at different locations within a cell is now captured more completely.

Apr 13

View all updates

User Base

50+ enterprise customers

Active Users

Security & Privacy

US (Ohio), EU (Ireland)

SOC 2 Type II GDPR Compliant HIPAA Compliant

End-to-end encryption (TLS 1.2+ in transit, AES-256 at rest) Zero Data Retention (ZDR) option Single Sign-On (SSO) via SAML 2.0 and OIDC Role-based access control (RBAC) Audit logs and activity monitoring