Case study · Salty Cowboy, Bali · Side project
Threshold, a booking engine for Salty Cowboy
How a horse rescue turned manual booking approvals into a system that only offers rides that fit.
- The customer
- Riders booking with Salty Cowboy, and the owner who runs it
- Team
- Just me: I designed, built and shipped it
- Skills
The problem
Every booking arrived as a WhatsApp message. The owner screened each rider by hand, in three languages, and every mismatch caught late was a slot that could sit empty.
How I solved it
- 01Found the real constraint: nothing on the market held bookings and horse welfare together, and the owner won’t do admin, so nothing could depend on them maintaining it.
- 02Ruled out a settings page and built every rule into the product itself: a rider weight cap, per-activity durations and pricing, and a three day course booked as one block.
- 03Asked riders for their experience and weight up front, so rides and horses that don’t suit them never appear.
- 04Made it trilingual (English, Indonesian and Russian), with a clean English summary the owner approves in one WhatsApp message.
- 05Used AI across all of it: discovery synthesis, the design system and the build.
Outcome
The owner still gives the final yes on every ride. They just no longer have to be the filter.
Inside the work
-
The rider enters their experience and weight up front.
-
Threshold rules out rides and horses that don’t suit them. Those options never appear.
-
The rider books only from what fits.
-
The owner approves a clean summary in one message.
What I learned
Salty Cowboy isn’t unusual. Any business where the right option depends on who’s asking runs the same loop, and Threshold is built for exactly that.