Twill Brain

Data Insights

2 min read

Data Insights

Date: 2026-03-19 · Duration: 63.130001068115234 min Organizer: Nader Attendees: Nader, Ahmad, Mike, Martin

Summary

  • New Partnership with ISO Layer: Focus on large ISO with 20,000 merchants, connecting to Thesis for complex data handling.
  • Residual Data Complexity: Need for transaction-level analysis; NDA and varying formats pose challenges; pilot testing expected to start soon.
  • Recommendation Engine Development: Rule-based system to match merchants with processors; metrics on risk and savings essential for decision-making.
  • Scalability of Architecture: Database scaling concerns based on transaction volume; legacy systems may require redesign for efficient processing.
  • Market Positioning Strategy: Aim to build a solutions engine for streamlined processing, reducing reliance on middlemen over 5-10 years.
  • Long-Term AI Operating System Vision: LLM’s potential as an OS for decision-making in underwriting and onboarding remains exploratory.

Action Items

  • Martin Elias Costa Create a minimalistic ticket to define the framework and triggers for processor recommendation logic and UI integration (60:40) Implement a proof-of-concept involving one LLM call to evaluate output and feasibility of LLM in recommendation logic (59:30) Review Mike’s knowledge base folder and assess how it aligns with existing architecture and logic requirements (32:15) Define hard rules from knowledge base to exclude processors based on merchant application fields and implement this logic for initial QA and frontend display (46:32)

Ahmadreza Abdoli Share Mike’s knowledge base folder link in Slack for team access (32:40) Progress the recommendation system ticket on own board to align with new framework discussions and complete remaining small parts (51:10)

Nader Atrchin Coordinate board review session with Martin to prioritize recommendation logic and architectural adjustments (61:15) Oversee integration planning for backend recalculation triggers and data processing pipeline per discussion (52:54) Continue facilitating communication regarding the AI-assisted onboarding and processor matching development progress (60:40)

Graph View

Backlinks