Omni introduction

Date: March 25, 2026 Source: Granola Meeting ID: 3b1f90a9-85d2-487e-8da5-31d33a231f0f URL: https://notes.granola.ai/t/3b1f90a9-85d2-487e-8da5-31d33a231f0f

Participants:


Omni Tool Overview & Demo

  • AI-powered BI tool replacing spreadsheet reporting with direct Snowflake queries
  • Features semantic layer with Element business context and metric definitions
    • Enforces guardrails vs raw LLM database queries
    • Live connection to Snowflake (refreshes every few minutes)
    • Cache layer configurable to prevent warehouse overload
  • Current pilot scope: retail and wholesale data only
  • Greg Stoutenburg (Brainforge) leading development on branch system before production merge

Demo Findings & Issues

  • AI assistant “Blobby” successfully caught revenue vs point-of-sales distinction
    • Requested clarification between net sales vs gross sales
    • Shows training on Element’s definitions working
  • Data accuracy problems identified:
    • Target POS showing ~11M vs actual ~6M (double-counting issue)
    • Chart including total + drink mix + sparkling rows simultaneously
    • Walmart/Target POS data mixing incorrectly
  • SQL visibility available through explore button on any dashboard/chart
  • Natural language chart editing and dashboard saving functionality working

Pilot Structure & Next Steps

  • Contained rollout through summer (not May launch as originally planned)
  • Focus areas for testing:
    1. Point-of-sales data and store inventory velocity
    2. Retail data QA vs existing spreadsheets
    3. SQL query validation by tech team
  • Slack feedback channel created for bug reports and feature requests
  • Current data limitations: no Element retail revenue (purchase orders) in system yet
  • Greg pushing branch to production for team access and testing