Data Insights

Date: 2025-10-01 · Duration: 85.4800033569336 min Organizer: Nader Attendees: Nader, Ahmad, Mike, Martin

Summary

AI classification capabilities evaluated using real merchant data indicated timeout issues with a 30-second limit causing upload failures.
Document classification is functioning for most standard types, but challenges remain with funding applications requiring focused enhancements.
Proposed LLM specialization approach aims to increase accuracy, moving away from the current monolithic system towards tailored solutions.
Specific extraction rules for transaction data processing statements are necessary to enhance overall document accuracy and reliability.
New partner setup with Chad for document processing and lead generation has been initiated, which includes integration plans for MCA Tracker.
EMS data integration efforts are still blocked despite multiple transaction attempts, indicating a need for further investigation.
Historical data processing is ready, but current month estimation is blocking commission calculations, requiring interim reporting solutions.
Manual report processing has been established as a temporary measure for unique commission calculation scenarios.
Enhancements to the document category for funding applications are underway to better align with existing systems in PayEngine.
Continued focus required on optimizing document classification to cover edge cases in addition to typical processing scenarios.

Ahmadreza Abdoli Create new document category for funding applications that maps to ‘other’ in PayEngine while maintaining internal ‘funding application’ classification (22:12)

Martin Elias Costa Investigate timeout issues with 30-second document upload limit and implement retry mechanism improvements (09:18)

Michael Wright Confirm new partner Chad’s email and ensure portal access after email confirmation (01:21:55)

Nader Atrchin Create detailed ticket for LLM extraction improvements including website scraping, MCC logic, and transaction mix rules (59:07)