Fairway Labs
fairway advisors · internal
Caleb’s playground for new apps, customer demonstrations, and the occasional wild idea.
Most of what you see here isn’t carrying production traffic yet — these are demos, concepts, and works in progress, on purpose. The pace has picked up lately: the App Launcher can take an idea, typed or spoken, and turn it into a live tile here once it’s approved; a shared brand engine lets a new demo pick up a client’s colors and fonts without writing new CSS; and CH Tee Time Sniper — built on Playwright driving a real browser — has already graduated from Concept to a public app running against live systems. The Pipeline Pace strip above is the real cadence, not a mockup.
When something’s ready to carry real traffic, it goes to Ryan for hardening — GCP monitoring and alerting, secrets kept in a real vault, VPC and network planning, and proper OAuth/OIDC in place of a shared token — the same standard we hold every client engagement to. See our privacy policy for how we handle data across everything shown here.
Pipeline pace
Public Apps
Real, hosted, live — used by real humans today.
CH Tee Time Sniper
Books the preferred slot the instant the club’s tee-sheet release window opens, now running on a dedicated always-on backend — live for real members today.
Headless browser polls the portal in the seconds before release, NTP-synced to cut human reaction time. One GCP VM runs three services: web server, booking worker, and a timed read-only health check. Booking intake, events, and lead capture now flow through a dedicated Cloud Run API backed by Postgres, replacing the original Sheets-based collector for these paths. Members can pull up the club’s own booking screen for any confirmed slot; in-app delete is in development.
Concept
Past the idea stage — a working demo exists, though it's not yet in active build.
Is It Good?
Concept-stage camera prototype exploring a fast, objective gimme-putt check — an early look at a possible future smart-glasses feature, not a tool being shipped for the Oct 7 tournament. Uses the known fixed diameters of a golf ball and cup as calibration references (no AR/LiDAR).
Tests the math behind a possible future smart-glasses feature — a fast, objective gimme-putt call — as an innovation-path exploration, not a dependency for Oct 7 (gimmes there are settled with a printed ruler).
Storm Alert — Pro Shop
Warns the Pro Shop the moment lightning strikes inside a configurable radius of Chatham Hills, so staff can clear golfers before a storm arrives.
Chatham Hills' Pro Shop currently relies on staff watching the sky during a storm. A radius-based strike alert makes that automatic and immediate instead of word-of-mouth.
Law Office Automation
A fourteen-module concept build for running a small law firm end to end — lead intake, case-portfolio and contract tracking, condemnation campaign automation, and firm-wide financial reporting, all under one brand-configurable shell.
Every module here is real UI on fabricated data — a rehearsal for the actual build, not a live Google Workspace or Gemini integration (that's still a hardening-notes item, not a working feature). Small firms lose real hours to the same slog: intake scattered across email and spreadsheets, case and contract status tracked from memory, billing reconciled after the fact instead of as it happens. This build treats intake through collections as one connected system under a single brand-configurable shell, so the same foundation can be re-skinned for a firm's own case load without asking them to adopt new software.
Idea stage
Sketches and back-of-napkin concepts. No commitments, no timelines.
Human vs Claude Math Race
To illustrate the performance of solving simple math problems, racing a human against Claude to compare speed and accuracy, and to show the probabilistic nature of AI-generated answers versus deterministic human calculation.
We want to have interesting, simple ideas for the idea to launch framework. To inform the audience on how Humans may be better at simple math due to the probabilistic nature of AI.
Currently in testing and gathering user feedback — targeting a September publish.