Origin Story
From planning frustration to a working product
Parcours didn't start with a pitch deck. It started with too many tabs open, conflicting forum advice, and a genuine question about whether better planning software was actually possible. This is what happened next — the real timeline, not the startup version.
Why this exists
Cycling trip planning sits in an awkward middle ground. There's no shortage of inspiration — YouTube, Instagram, forums, route-sharing platforms — but very little that helps a rider make the specific decisions that actually determine whether a trip goes well.
The gap isn't information. It's judgement. Which destination fits your fitness? Which base town minimises transfer friction? How do you structure days that are ambitious without being reckless? These are the decisions that planning tools largely don't help with — they either show routes, or they show inspiration, but they don't reason.
The timeline
Before 2024
Trip planning was already part of the enjoyment
Long before this was a product, researching a cycling trip was a ritual in itself — comparing regions, mapping out cols, working out logistics. The research was time-consuming, but it was also part of the anticipation. No one wanted to skip it.
2024
One multi-location trip took roughly two months to plan part-time
A trip with three bases across two countries involved spreadsheets, scattered notes, forum threads, maps, and a lot of back-and-forth. Not because the information was hard to find, but because no single tool held it together or helped weigh the real tradeoffs.
2024
ChatGPT helped — but it wasn't the engine
Used for drafting logistics lists, sanity-checking route density, and bouncing itinerary ideas around. Helpful at the margins, but not reliable enough to trust for trip decisions. The context window was too short, the knowledge too broad, and the outputs too inconsistent to build anything around.
2025
An injury created space to experiment with Custom GPTs
Time off the bike meant time to test whether structured AI tools could actually reduce planning friction. Custom GPTs were a useful sandbox — constrained enough to be somewhat reliable, flexible enough to test different planning flows without a full build.
2025
A proof of concept existed, but constraints blocked the path to an MVP
The GPT experiments showed something worth building. But the Custom GPT platform had hard limits: no persistent context across sessions, no structured data access, no real planning state. The gap between proof-of-concept and shippable product was significant.
2026
Vibe coding with Claude and OpenClaw made a functional product possible
What changed was the workflow, not the ambition. Rapid prototyping with AI-assisted development removed the bottleneck between idea and working software. The constraint shifted from "can we build it?" to "does this actually solve the planning problem?" — a much better question to be solving.
Start from a real planning question
Origin stories are easy to over-tell. The thing that matters is whether the product actually helps — whether it makes your next trip easier to plan and more likely to go well. That's what Parcours is for.
This is not a startup version of the origin story. This is what actually happened.