Things I've shipped with Claude. Mostly to make my own job easier.
I don't write production code from scratch. I ship products with Claude. Eight of them so far, across web, mobile, internal tooling, methodology, reporting, a public calculator, and an indicator I run on my own portfolio.
Most marketers can audit a website. Fewer can ship one. I work with Claude as the engineering hand and bring the product judgment, taste, and SEO instinct. Listed because the shape of these says more about how I work than any case study.
This site: drewgarrett.org.
Personal portfolio replacing a legacy WordPress + Elementor build. Hand-coded in Astro 5 + Tailwind v4 + TypeScript on Vercel. Every line shipped with Claude Code as the engineering tool.
Drew's Average Day: a playable browser game.
A small pixel-art day-in-the-life game built into this portfolio. Walk the neighborhood, lift, work, clean, eat, and choose when to call it a night—with every decision shaping the final score.
Programmatic SEO audit methodology.
Built with Claude on a creator-tools SaaS account: a weighted scoring model (0–100) that audited ~929 programmatic pages, bucketed them into Keep / Improve / Consolidate / Kill, mapped 45 topic-based hubs, and flagged 138 cannibalization risks. Self-initiated; reusable across any account with programmatic-page sprawl.
Looker Studio dashboards and reports.
Recurring reporting dashboards built across every account: GSC and GA4 performance, keyword tracking, and the newer LLM-visibility layer pulling from Scrunch AI. The standard client-reporting tool for most of this work.
Honestly, I don't think Looker is necessarily the best way to build most reports anymore. AI can query and summarize the same raw data directly, on demand, without a dashboard to maintain. Building these the old way is exactly what makes that gap obvious.
SEO ROI calculator.
Every SEO ROI calculator online multiplies sessions by a conversion rate by a deal value and prints a flattering number. This one reports gross profit rather than revenue, uses lifetime value spread across the customer's actual life instead of booking it all in the signing month, and simulates break-even month by month as cohorts accumulate. On the default scenario that means year one comes back negative and break-even lands in month 14, which is the honest shape of an SEO program and the reason no vendor calculator shows it. Ungated, no email, and the inputs live in the URL so a scenario can be sent to a CFO.
SEO content brief generator.
An internal SEO tool I built and use at the start of every engagement. Spotted a recurring workflow that could run faster and more consistently. Coded the fix, folded it into how I work.
Align: iOS habit + workout app.
Personal habit and workout tracking app, shipped to my own device. Built entirely with Claude Code in Swift / SwiftUI.
Custom trading indicator.
A trading indicator I built in Pine Editor on TradingView. Maps my own framework for weekly calls on stocks I'm bullish on into buy / sell entry signals, and pushes alerts to my phone, so I don't have to stare at charts all day.
Why these exist at all.
None of these are products. The making is the thing. Each one exists because I spotted a recurring task that could run more efficiently or more consistently, and decided to fix it instead of doing it the slow way again.
Same heuristic I apply on retainer at Directive. The build is usually small; the harder part is noticing what's worth tooling for, and then choosing to do something about it.
Claude does the engineering. I bring product judgment, taste, and SEO instinct. I'm a marketer who ships software, not a developer who wandered into marketing. Building helps me understand what's actually possible.