Twill Brain

Data Insights

3 min read

Data Insights

Date: 2026-03-04 · Duration: 85.12999725341797 min Organizer: Nader Attendees: Nader, Ahmad, Mike, Martin

Summary

  • AI Onboarding Agent Design: Building an AI chat agent for guiding application completion and processor selection through interactions.
  • Role-Based Views: Different views for Merchant, Partner, and Master users provide tailored tools and information visibility.
  • Human in the Loop: AI escalates to humans when necessary, using intent classification for efficient routing of complex queries.
  • Chat Interface: Main interaction occurs through chat, with notifications and actionable widgets reducing the need for dashboards.
  • Event-Driven AI: AI responds to application status changes, suggesting actions proactively using event-driven triggers for real-time updates.
  • Document Handling: Structured templates for document types enhance data accuracy and support conversational requests for missing fields.

Action Items

  • Nader Atrchin Share the AI onboarding agent specification document in Confluence for detailed team review (03:43) Refine AI onboarding agent spec to include realistic onboarding questions and align with discussed workflows (05:54) Coordinate with Morty AI to generate tickets based on the agreed spec, including test criteria and acceptance conditions for each tool and epic (19:17) Enhance Morty AI capabilities on GitHub to assist with PR attachments and code generation independently (01:03:12) Determine the visibility settings for AI widgets and controls to restrict merchant and partner views appropriately (01:17:17) Expand structured document processing logic to assign specific extraction templates per document type (01:18:26) Adjust the onboarding agent UX to prioritize widgets with associated actionable next steps, removing passive information displays (38:56) Continue iterative brainstorming on event-driven architecture and websocket event triggers for AI responsiveness during onboarding updates (01:05:36) Schedule and lead follow-up meetings to continue discussing design refinements, ensuring everyone has access to key resources (01:24:22)

Martin Elias Costa Provide feedback on the onboarding conversations and suggest realistic onboarding queries from merchant and rep perspectives (05:20) Review and critique widget designs, emphasizing action-oriented notifications and minimizing information-only clutter (30:01) Assist in defining the separation of next best action engine and AI chat assistant roles as distinct but interacting systems (38:56) Collaborate on organizing AI chat sessions with robust context management and support for parallel conversations (55:39) Advise on keeping UI interactions simple and clear, favoring direct action over complex AI command chains (49:27)

Michael Wright Advise on solution-related AI tools, including processor eligibility and deal recommendations integrated into onboarding agent flows (13:07) Provide insights on mapping solutions and offers to merchant onboarding paths (24:52) Contribute to defining the classification of AI intents and human-in-the-loop triggers for handling merchant requests (01:15:02) Update the team on CRM vendor discussions integrating AI features similar to chat-driven database queries (45:11)

Graph View

Backlinks