Abstract
Verification of document packets for regulatory compliance may rely on manual review, which can be inefficient and susceptible to error. Some automated tools may not adequately compare data across diverse structured forms and unstructured evidence within a packet. A multi-agent artificial intelligence system can use specialized agents, which may leverage large language models, to automate processes such as the ingestion, classification, extraction, and validation of information. The system can employ a configurable source-of-truth hierarchy and multi-tiered matching logic to validate data. A human-in-the-loop mechanism can present flagged discrepancies or unverified fields to a human specialist for review and adjudication. This approach may improve the efficiency, consistency, and auditable governance of high-volume document verification workflows.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Sankar Sikakolli, Bhavani; Konidala, Kalyan; Badjatiya, Pinkesh; Indoria, Yogesh; Ghosh, Gourabdip; Kurmi, Siddhant; Singh, Aryan; Basu, Tilottama; Rawani, Farhan; and Mohite, Sangram, "Multi-Agent AI System for Human-in-the-Loop Document Packet Verification", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10692