"Agentic Document Extraction Upgrade for Complex Tables"

Announcing a significant upgrade to Agentic Document Extraction! LandingAI's new DPT (Document Pre-trained Transformer) accurately extracts even from complex docs. For example, from large, complex tables, which is important for many finance and healthcare applications. And a new SDK makes using it require only 3 simple lines of code. Please see the video for technical details. I hope this unlocks a lot of value from the "dark data" currently stuck in PDF files, and that you'll build something cool with this!

🚀 Big News: Agentic Document Extraction (ADE) just got a major upgrade. We’re introducing Document Pre-trained Transformer 2 (DPT-2), a new foundation model that powers the next generation of ADE. 🔍 Why this matters: Parsing complex, messy documents has always been error-prone. Tables without gridlines, scanned invoices at odd angles, embedded stamps or signatures — traditional systems miss them. DPT-2 brings higher accuracy, faster performance, and full grounding to every extraction. 💡 What’s improved: • Parse large, no-gridline tables cell by cell with precise alignment • Smarter layout detection in messy scans with fewer missed chunks • Expanded coverage for signatures, checkboxes, barcodes, and QR codes • Concise figure captioning for logos and seals without verbose noise • Parallel extraction for developers via API and SDKs • Reliability for industries where accuracy is critical: finance, healthcare, insurance, compliance Read to see it in action? 👉 Try ADE DPT-2 in the Playground and API: https://lnkd.in/eZq-NqWH

Great to see this landmark upgrade to the agentic AI framework in action! Andrew Ng’s vision - moving beyond linear prompt/response models and toward dynamic, autonomous workflows - continues to redefine what’s possible for enterprise AI and digital transformation. Agentic architectures and orchestration layers are turning large language models into true collaborators, empowering AI agents to plan, reflect, and iterate in service of real-world business goals. The results: higher quality outputs, faster prototyping, and the ability to deploy applications that learn and evolve with each task. This upgrade is a game-changer for AI accessibility and value creation. At ACL Digital, we are looking forward to seeing organizations everywhere harness the full spectrum of agentic capabilities to drive innovation and deliver impact at scale! #AI #AgenticAI #Innovation

A new era of intelligence is unfolding as LandingAI’s upgraded Agentic Document Extraction with DPT transforms the way we unlock value from complex, unstructured data. 📑⚡ From intricate financial tables to dense healthcare records, this breakthrough makes extraction seamless—while a powerful SDK reduces integration to just three lines of code. By liberating “dark data” from static PDFs, this innovation opens boundless opportunities for smarter, faster, and more impactful solutions. 💚 #AI #DarkData #Innovation #DataExtraction #FutureOfWork #LandingAI

>> document pre-trained transformer model by LandingAI . The Agentic Document Extraction converts complex documents with the embedded rows and charts into LLM-ready data. It takes just a few lines of code to use the SDK building the powerful #Agentic #RAG answer #engine based on the existing docs. link https://lnkd.in/d52Bpx8N

Just experimented with LandingAI’s new Agentic Document Extraction powered by DPT - and I’m genuinely impressed. I revisited our Capstone project’s quarterly earnings presentation (a 50-page PDF packed with charts and tables), and it parsed everything in under 20 seconds with remarkable accuracy. Having experimented with multiple libraries for this task - each with its own quirks - this release feels like a leap forward in document intelligence. A brilliant tool for anyone working with complex visual data. I highly recommend giving it a spin: https://va.landing.ai/

💡 Great to see advancements like DPT pushing the boundaries of document intelligence. CNNs have already shown strong capabilities in extracting information from complex layouts and even handwritten notes, reducing the time spent combing through massive volumes of unstructured data. 🎢 What excites me is how combining CNN-based approaches with transformers can further minimize information loss, especially in domains like healthcare and finance where every detail matters. This really opens the door to unlocking the “dark data” hidden in PDFs and scanned docs. What's your thoughts on this? #LandingAI #ML #CNN #AIResearcher

This is very cool...We've all, well most of us that work in business, have run into this problem... Here is a tool that can make extracting data from structured and unstructured data and file type... but you don't have to take my word for it... here's Andrew Ng to explain. Thought I would share because this is a common use case and could help you build something awesome..

This is a significant boon for anyone working with documents where traditional OCR can’t understand complex tables with merged sections areas embedded in PDFs. Often these get misinterpretted in traditional RAG injestion and end up incorrectly identifying the relationship’s, an then providing inaccurate data in return. This will bring significant improvement’s in enterprise’s search and finacial data indexing.

Interesting post as table transformation was an issue with the past LLM models , but there has been significant strides in tech . And we have leveraged it , now we can process any sort of RA sheets with any format with these tools to bulk upload actions onto ehs tools