A side project · 2024 – 2026
EarliBirdi was a tee-time alert service I designed, built, and ran on my own. You told it where and when you wanted to play. It watched the course's booking system around the clock and sent you a text with a booking link the moment a matching time appeared. I've since retired it. This page covers what it did, how it worked, and why I shut it down.
Good tee times at popular public courses go fast and cancellations come back without warning. EarliBirdi watched the tee sheet so golfers didn't have to refresh it all day.
Search by zip or city, pick a supported course, then set the date, time window, group size and holes.
A scheduled engine checked every active alert against the course's live tee sheet across three different booking platforms.
When a matching time opened up, an SMS with the course, time, price and a direct booking link went out right away.
I did all of it: product, design, engineering, payments, infrastructure and support.
A React web app took the alerts. A scheduled engine in Google Cloud did the watching. Twilio delivered the texts.
React 18 · Material UI · Firebase (Auth, Firestore, Functions, Hosting) · Cloud Run · Cloud Scheduler · Secret Manager · Twilio · Stripe · GitHub Actions
An alert that quietly never fires looks exactly like one with nothing to report. I built a validator that ran the engine's exact query against every course and flagged any that went dark, plus a heartbeat that paged me if a sweep didn't run.
Each booking platform had its own API, IDs and failure modes. A provider-adapter layer normalized them, so adding a course became a one-line data change instead of new code.
The engine started as a loop on a computer in my house. I moved it to Cloud Run Jobs on a schedule, with stateless sweeps, secrets in Secret Manager and per-commit images for one-command rollback.
Once it took payments, I audited it myself. I moved plan limits server-side, locked down database rules so no one could upgrade themselves to Pro, and normalized phone numbers so every alert reached a real phone.
The alert arrives as a text, so you'd act on it from your phone. Every flow was designed for mobile first.
EarliBirdi did what it promised. It found tee times and sent the texts, and building it taught me more about running a product than anything I could have read. I retired it on purpose, for two reasons.
The platforms caught up. When I started, booking systems treated last-minute openings as an afterthought. Over the next two seasons that changed. Courses and their booking platforms launched their own waitlists and notifications, and started restricting automated access from third-party tools. One platform put its entire tee sheet behind a CAPTCHA, and others changed their APIs without notice. The problem I set out to solve was being solved at the source, which is the best validation an idea can get.
That changed the job. Keeping alerts accurate stopped being a product problem and became a permanent integration problem: chasing undocumented changes and access rules that were never built with a partner in mind. I wasn't going to engineer around a platform's decision to limit access. An alert service that misses alerts is worse than none at all.
So rather than let it fade, I shut it down deliberately.
And my full-time work deserves my full attention. In 2026 I joined Oasis Security, now part of Cyera, where I'm a Senior Customer Success Engineer. EarliBirdi had grown into a second job. I'd rather do one job exceptionally well than two adequately. So the alerts stopped, user data was deleted, and this page replaced the app.