In this episode of Ride with Joe, Joe Maraschiello discusses a new framework he developed called BAPA (Brand AI Perception Analytics), which focuses on the emerging field of Generative Engine Optimization (GEO). The goal of BAPA is to help brands understand how they are perceived by AI agents (like ChatGPT and Claude) and to move AI workflows from complex, code-heavy prototypes into sustainable production environments.
Key Pillars of the BAPA Framework:
- Visual Workflows: Instead of relying on complex code, Joe advocates for using tools like KNIME to create point-and-click visual workflows (09:35). This improves transparency, makes auditing easier, and allows non-technical business users to understand and maintain AI processes (21:32).
- LLM Agnostic Design: By using visual nodes, the framework remains flexible, allowing users to swap out different LLMs as the technology evolves without needing to rewrite entire pipelines (24:23).
- Medallion Architecture: To ensure data quality and collaboration with data engineering teams, BAPA uses a medallion-based architecture (Bronze for raw data, Silver for enrichment/sentiment analysis, and Gold for final analytics) and outputs the results to BigQuery for integration with existing digital analytics infrastructure (15:47).
Why BAPA Matters:
Joe emphasizes that current SaaS solutions for brand visibility often lock users into black-box systems, while custom open-source scripts are often too technical for business teams to manage long-term (06:09). By leveraging visual design, he aims to “humanize” the AI workflow, ensuring that businesses can stay agile in a fast-changing environment while maintaining control over their data (23:46).