AI Dev 25 X NYC | David Park: Impact of Agentic AI in Financial Services on Document Extraction

Why It Matters

ADE lets financial institutions automate and audit complex document workflows, dramatically cutting manual effort and regulatory exposure while accelerating loan and compliance processes.

Summary

David Park, head of Applied AI Engineering at Landing AI, introduced the company’s new Agentic Document Extraction (ADE) platform, positioning it as a developer‑first, enterprise‑grade solution designed to modernize multimodal document processing for financial services.

He detailed ADE’s three‑tier architecture: a foundation layer of document‑pre‑trained transformers that understand complex layouts and mixed content; a parsing layer that converts visual and semantic insights into structured JSON/Markdown with layout awareness; and an agents‑and‑apps layer that delivers field extraction, classification and workflow automation, all while supporting flexible deployment models and strict compliance standards.

Park highlighted a real‑world KYC deployment with a major bank that slashed manual document review by over 70%, and illustrated a loan‑packet use case where ADE automatically classified, split, and extracted fields from heterogeneous documents—pay stubs, W‑2s, IDs—providing traceable, audit‑ready outputs. He also showcased the platform’s APIs for schema‑driven extraction and a chat‑based playground that lets users query extracted data with cell‑level provenance.

The platform promises to reshape financial compliance and lending operations by reducing labor‑intensive review, lowering regulatory risk, and enabling rapid iteration of AI‑driven workflows, giving institutions a scalable edge in a data‑heavy, highly regulated environment.

Original Description

David Park, Director of Customer Facing MLE at Landing AI, focused his AI Dev 25 x NYC talk in financial services on multimodal document extraction.

Agentic AI is transforming financial services by enabling intelligent, autonomous workflows that combine language, vision, and structured reasoning to extract insights from complex, multimodal documents at scale.

By orchestrating specialized models and decision-making agents, these systems move beyond static OCR and rule-based parsing to handle diverse data types — from regulatory filings and KYC forms to contracts and blueprints — with greater accuracy, context awareness, and adaptability.

This shift unlocks new possibilities for automation, compliance, and real-time decision intelligence across the financial ecosystem.

----------------

Join us at AI Dev 26 x San Francisco! Tickets: https://ai-dev.deeplearning.ai/