LandingAI

Introducing Split Classification in ADE - Content-aware document splitting for multi-document PDFs 🚀 Many customers receive large PDFs that aren’t a single document at all. They’re bundles: intake forms attached to clinical notes, invoices mixed with authorizations, packets where every page belongs to something different. Now once you upload and Parse the document using ADE, imagine running Extract on that entire file. You end up processing pages that don’t belong together, applying the wrong schema to the wrong section, and spending credits on content that isn’t even relevant to the task. Before any Extract or downstream workflow can make sense, the file needs to be separated into the right document groups. ADE Split solves exactly that. It analyzes a large file and returns structured groups of pages, each representing a single document type. The PDF is never modified. The API returns clean JSON that includes: • Document class suggestions • Optional identifiers such as patient name or invoice ID • Page ranges that define each grouped document • Markdown for the identified sections • A required “uncategorized” group for pages that don’t match any class The best part? The grouping is context-aware. Pages without explicit identifiers still cluster based on content, and a single page is never assigned to more than one document. ADE Split is now available in preview in both the Playground and the API, and works with DPT-2 and DPT-2 Mini. ➡️ Try it out here: https://va.landing.ai/ Read more: https://lnkd.in/gNJM5QzJ

Live Webinar - Scaling Healthcare RCM with Agentic Document Extraction! Healthcare RCM teams deal with some of the hardest document formats: claims, EOBs, medical records, appeals, and dense multi-page PDFs. In this session, we’ll show how ADE handles these layouts at scale without templates, hand-built rules, or model tuning. 📆 Dec 10th - 9 AM PT 🔗 Register here: https://lnkd.in/gsbN9hV7

Managing OTC derivatives collateral shouldn’t feel like a full-time job. One of the standout projects from our Financial AI Hackathon and winner of the Best Online Team Award tackled this head-on. If you’ve ever touched ISDA/CSA agreements, you know the pain: PDFs, messy tables, inconsistent formats, manual margin calculations, and auditors who want everything traced back to the clause and page number. The team built a Margin Collateral Agent that automates the full workflow of collateral management under ISDA/CSA agreements. What it does: • Extracts CSA terms directly from PDFs using LandingAI Agentic Document Extraction(ADE) • Normalizes collateral tables with an AI-driven workflow • Calculates margin using a deterministic rules engine • Generates plain-language explanations with clause-level citations • Maintains a complete audit trail for compliance The flow stays simple and clean: Upload → Parse → Extract → Normalize → Calculate → Explain Why this matters: Margin calls depend on accuracy. Humans are slow and make mistakes. Regulators do not care about excuses. This agent gives you speed, determinism, and transparency in one place. Full demo and project link in the comments!

Native-Language Document Extraction with ADE! Most extraction tools depend on translation pipelines, which introduce loss of meaning, misclassification, and extra steps for teams working with non-English documents. This blog breaks down why this “translation tax” creates accuracy gaps in medical, financial, legal, and administrative workflows. Agentic Document Extraction (ADE) removes that dependency by supporting complete native-language processing from input to output. The blog walks through how ADE enables: • Direct parsing of non-English text • Schema definitions written in natural, domain-specific language • Structured output in the original language • Higher precision on technical and operational terms • Zero translation overhead for users The post also highlights real use cases, including health record analysis, financial document verification, and employment workflow checks, showing how native-language extraction improves accuracy and reduces manual review. Link to the blog in the comments!

We’re hosting this week’s ADE Office Hours on Thursday. ⚡ We recently launched DPT-2 mini, our lightweight model for high-volume, predictable, digitally generated documents, and the response from the community has been incredible. If you need help integrating or evaluating DPT-2 mini for your use case, or if you're working with ADE in general, bring your questions and join us for direct guidance from the team. This is an open hour with Ron Baumert, Ava Xia, Seshu Reddy, and Emilie Cooksey. 📅 Thursday, December 4 ⏰ 9:00 AM PT 👉 Register here: https://lnkd.in/g5Qg-N7f

Introducing DocuFlow - Document intelligence for finance teams! DocuFlow won second prize at the Financial AI Hackathon. 🥈 The team built an end-to-end document intelligence system that helps analysts work through invoices and contracts with fast extraction, search, and review tools. Why this matters: Finance teams spend a lot of time scanning through PDFs to find amounts, clauses, mismatches, and compliance issues. DocuFlow streamlines this by automating extraction, search, and inspection, so analysts can move from document upload to investigation without switching tools. Here is the complete workflow: • Upload documents through a simple interface • Parse and extract fields using LandingAI's Agentic Document Extraction (ADE) • Generate embeddings for semantic search and RAG • Store extracted fields, metadata, and vector embeddings in PostgreSQL with pgvector • Review results in an interface that supports search, extraction display, RAG-based answers, and compliance checks A strong example of a practical document workflow: clear extraction, fast retrieval, and a reviewer experience that keeps analysts inside one system. Full project and demo in the comments!

Automate finding Best Black Friday Deals for your Budget using Agentic Document Extraction

Black Friday ads are chaotic. ADE turns them into a clean shopping list. Holiday ad booklets are packed with dense layouts, mixed fonts, and cluttered product grids. Great for deals, terrible for quick decisions. We ran a physical Black Friday booklet through LandingAI’s Agentic Document Extraction (ADE) and turned it into a structured product and price list in under a minute. Here is the workflow: 📸 Capture: Snap a photo of the busiest ad page. ⚡ Parse & Extract: ADE pulls out product names, prices, and details with no manual entry. 💬 Chat: Ask questions directly over the extracted data. Example: “Find the top 3 best value deals under $20.” Instant shortlist. No scanning through grids. Clear decisions. Watch how we turned paper heavy ads into a smart shopping strategy in seconds.👇 Try it in the Playground: https://lnkd.in/eZq-NqWH