VBIG is a prototype system that explores how small Vision-Language Models (≤3B parameters) can augment business intelligence workflows by generating concise, human-readable insights from BI dashboards.
The system combines:
- Vision-based insight generation (VLMs)
- Agentic orchestration (via n8n)
- Fact-check verification (through SQL access to BI backend tables)
- Supabase vector storage (dashboards + historical insights)
- Low-data-literacy support
- Daily dashboard simulation
VBIG serves as the applied component of a broader goal:
Embedding lightweight AI assistants into enterprise BI systems to improve clarity, speed, and decision support.