Three months into 2026, the most relatable AI project might be a Guinness price tracker.
A developer built a tool using Anthropic's Claude to monitor pint prices at pubs across Ireland - using AI to handle the messy work of collecting, parsing, and comparing pricing data from different sources, as covered by Fortune. The model processes pricing information in whatever format it arrives - text menus, social posts, inconsistent listings - and outputs something clean and comparable.
The appeal is obvious if you've ever paid €8 for a pint in central Dublin. Guinness prices vary significantly by neighborhood and establishment across Ireland, and locals pay attention. Rather than manually checking menus or relying on crowd-sourced apps, the builder used Claude as the data layer.
It's not a technically complex project, and that's the point. Claude handles the unstructured parsing work so the builder didn't need to write custom code for every new data source. No elaborate pipeline, no database schema upfront - just a model that can read messy inputs and return structured comparisons.
This is how most people actually use AI tools day-to-day: not building enterprise products, not automating complex workflows - just solving a specific personal problem faster than a spreadsheet would allow. The Guinness tracker is a small project, but it's a more honest illustration of practical AI use than most case studies you'll read.