Case Studies

A researcher was spending weeks on each data cycle - pulling records, reformatting spreadsheets, cross-referencing entries. The same tedious steps, every time. Nobody had questioned it.

A custom tool that automated the extraction, transformation, and cross-referencing workflow. Used AI for the parts that required judgment - pattern matching, anomaly flagging - and straightforward automation for the rest.

What took weeks now takes minutes. The researcher said it saved her months of cumulative work. The approach was validated for expansion to other manual workflows in the organization.

With solar panels, a battery, and TOU rates, every night is the same question: how much battery to charge from the grid while it's cheap? Charge too much and you waste money. Too little and you buy at peak rates. Nobody had automated this for Home Assistant.

A full HACS integration with three-tier rate awareness, solar forecast simulation, off-peak battery hold logic for winter months, automation blueprints, and push notifications. Inverter-agnostic - works with any battery system.

11 releases, 2 contributors, shared with the DIY Solar Power Forum community. Runs nightly on my own system managing 23+ kWh of battery across two properties.

Home Assistant tracks energy consumption but not energy cost. For TOU customers, knowing you used 10 kWh is meaningless without knowing which rate tier it fell in. This gap meant people couldn't see the financial impact of their usage patterns.

A HACS integration that multiplies consumption by the correct rate at the correct time - with seasonal schedules, holiday recognition, and daily/weekly/monthly cost sensors. Built a companion rate calculator to help people figure out their actual effective rate from their utility bill.

8 releases, paired with the JSS rate calculator. Gives homeowners the real-dollar picture of their energy decisions.

Existing AI code review tools are bland. They give you the same polite, sanitized feedback every time. Sometimes you want someone to actually roast your code. Sometimes you want a security expert, not a generalist.

Built the first version in an evening for fun. Then iterated: 8 reviewer personas from "Best Practices Guru" to "Meanest," model selection between Haiku and Sonnet, rate limiting, input validation, serverless Cloudflare architecture. 37 commits total.

Fully deployed on Cloudflare Pages with auto-deploy on push. A working demonstration that an idea can go from frustration to production-grade tool quickly when you know the stack.

PGE's residential rate structure is deliberately opaque - generation charges, transmission, distribution, fixed fees, taxes, all broken out differently. Your actual effective rate is about $0.211/kWh, but try figuring that out from your bill. Nobody was giving homeowners a straight answer, and you can't evaluate solar ROI without one.

A rate calculator that takes your actual bill data and computes your real effective rate. Separates volumetric charges from fixed costs. Outputs the exact numbers to plug into Home Assistant for accurate energy cost tracking.

Part of the Johnny Solarseed project - a full educational ecosystem for Oregon DIY solar homeowners. The calculator feeds directly into the TOU metering HACS plugin.