The idea seemed straightforward: take a photo of those lengthy school supply lists that come home every August, use AI to extract and identify each item, then automatically search across major retailers to find the best prices. Parents could save hours of comparison shopping and potentially hundreds of dollars by finding the optimal combination of stores and deals.
We started with optical character recognition (OCR) to read the supply lists, then planned to use natural language processing to standardize item names across different retailers. The AI component would handle variations like "24-count Crayola crayons" vs "Crayons, 24 pk, Crayola brand" to ensure we were comparing apples to apples. From a technical standpoint, this part was entirely achievable with current AI tools.
That's where things got complicated. We tried to integrate with the top-5 stores - Staples, Amazon, Walmart, Target, and Office Depot - but found that each retailer severely restricts real-time pricing access. Amazon's Product Advertising API has strict approval processes and usage limits that favor established businesses. Walmart's API requires enterprise partnerships that are simply out of reach for small developers. Most retailers either don't offer public pricing APIs at all or charge prohibitive fees for access.
We attempted to leverage Google Shopping's API, thinking it would aggregate pricing across retailers and solve our integration headaches. However, Google Shopping charges enterprise-level fees for their comparison shopping service API - we're talking thousands of dollars per month in usage costs, plus strict commercial licensing requirements that put it well out of reach for a mom & pop operation like ours.
The harsh truth we discovered is that real-time retail pricing data is treated as premium commercial intelligence. The big players have exclusive deals, enterprise partnerships, and massive budgets for data access. There are hardly any ways to do this at the mom & pop scale without either violating terms of service or paying enterprise-level fees that would make the project financially impossible before it even starts.
This project taught us valuable lessons about the difference between "technically possible" and "practically feasible" for small businesses. While the AI and automation pieces were entirely doable with our skills and available tools, the data access layer proved to be the real barrier. It's a perfect example of how innovation at small scale often hits walls that have nothing to do with technical capability and everything to do with business gatekeeping.
We haven't given up on the core idea, though. We're exploring alternative approaches like crowd-sourced pricing, partnerships with smaller retailers who might be more open to collaboration, or focusing on specific product categories where pricing data is more accessible. It would still be a really cool product, and sometimes the best innovations come from finding clever workarounds to seemingly impossible problems.
This experience reinforced our belief that there's still plenty of room for innovation at the small business level - you just have to be more creative about finding paths that the big players haven't already locked down. Sometimes the most interesting solutions come from working around limitations rather than having unlimited resources.