Practical AI
Turning Internal AI Infrastructure Into Client-Facing Intelligence
As Head of Product and CEO at Evidnt, I led the development of our AI-powered classification and insights engine — a core capability that helped transform fragmented retail data into decision-ready intelligence for marketers, retailers, and CPG brands.
As Head of Product and CEO at Evidnt, I led a dual-track effort: first, to solve our internal data infrastructure challenges using AI — and then to turn that foundation into a scalable insights engine for brands and marketers.
The Internal Challenge: Scaling Product Classification
Before we could offer insight, we had to create structure.
Retail data comes in messy: inconsistent brand names, mismatched categories, duplicated SKUs, missing attributes. With over 1.2 million products flowing through our system — and growing — manual or rules-based classification simply couldn’t keep up.
We needed to:
Normalize products across 10,000+ brands
Organize data into 2,000+ category trees and 5,000+ unique categories
Assign flavor, package type, dietary, and size attributes to each SKU
Ensure accuracy, consistency, and scalability across 28,000 U.S. retail locations
We strategically built a proprietary AI classification engine that combined GPT-based large language models, machine learning, and proprietary validation logic. This allowed us to process, enrich, and structure CPG data at scale — turning a back-end bottleneck into a strength.
The External Solution: Real-Time Insights for Clients
Once our internal system was in place, we realized it could power much more than internal reporting.
We layered on a natural language insights interface that allowed brand managers to ask real business questions — and get answers instantly:
“Which flavor segments are growing fastest in the Midwest?”
“Where are my competitors gaining share in the 16 oz category?”
“What categories are over-indexing in my top DMAs?”
This became a differentiator, transforming our platform from a measurement tool into a real-time intelligence engine.
We integrated this system directly into The Trade Desk, Meta, Microsoft and Google, enabling smarter targeting, in-flight optimization, and media measurement based on actual retail sales.
The Results
What began as a strategic internal AI investment became the engine behind our client-facing platform:
Classified and enriched 1.2M+ products with structured brand and category data
Enabled insights across 28,000+ retail locations in near real-time
Supported both Fortune 500 and emerging brands in campaign planning, targeting, and optimization
Delivered double-digit sales lift through better audience and media decisions
Became the core intelligence layer driving measurement, activation, and strategy
We enabled brands to gain real time sales trends and inisghts utilizing AI based data querying.
My Role
I led the vision, architecture, and execution:
Defined the product roadmap and system requirements
Oversaw the AI model integration, scoring, and human feedback loops
Partnered with engineering and data science teams to build a scalable infrastructure
Collaborated directly with clients to refine use cases and embed insights into their workflows
This is what I believe AI should be:
Strategic, structural, and practical — driving business outcomes, not just automation.
