Webcam Taxi
A sceptical dev team, a tight deadline, and a project shipped two weeks early
Michael Heavey, Founder
"I'd been trying to convince my developers to try AI tools for a while with no luck. After the Sprint, they were set up on Claude Code and could see exactly how it fitted into the way they already work. The airport project went live two weeks ahead of schedule — that's not something that would have happened before."
Placeholder — awaiting Michael's quote.
Michael Heavey
Founder, Webcam Taxi
Background
Webcam Taxi is an online resource for live webcams at destinations around the world: airports, cities, beaches, ski resorts, landmarks. Founded by Michael Heavey, the site has been running for years and is maintained by a two-person development team who have worked with Michael for two decades.
Michael is a firm believer in AI and its potential to change how work gets done. His developers are experienced, capable, and deeply familiar with the codebase. But they were sceptical of AI coding tools, felt they didn't need them, and hadn't tried them. With a new airport webcam project on the horizon and a tight commercial deadline attached to it, Michael wanted that to change.
The challenge
The goal wasn't just to introduce a new tool. It was to change how an established, experienced team worked — people who had built their own reliable processes over two decades and had no obvious reason to adopt something unfamiliar.
Michael's underlying business goal made the stakes concrete: the airport webcam project needed to ship on a tight deadline to start driving revenue. That meant the Sprint had to accomplish two things at once: get the developers genuinely set up and confident on Claude Code, and map out how it could accelerate the airport project specifically.
The target was a five times or greater improvement in workflow speed, with repeatable processes and scaffolding the team could use beyond this single project.
What the Sprint covered
Getting the team set up properly
Both developers were set up on Claude Code during the session, with the mental model, environment, and working approach they'd need to use it effectively. The framing wasn't "replace how you work" but "fit this into what you already do well." For developers with deep existing knowledge of the codebase, the goal was to show Claude Code as a force multiplier on that knowledge, not a replacement for it.
Mapping Claude Code to their existing workflow
Rather than prescribing a generic process, the Sprint worked through how Claude Code could fit the specific tasks these developers were already doing: writing and reviewing code, building new features, debugging, and maintaining a large existing codebase. The output was a repeatable workflow they could apply immediately to the airport project and continue using across future work.
Scaffolding for the airport webcam project
With the airport project as a concrete, high-stakes test case, the session mapped out how to use Claude Code to accelerate the build: from scoping and architecture through to implementation and review. The developers left with the context and approach needed to move fast without cutting corners on a codebase they knew well.
What happened next
Both developers incorporated Claude Code into their daily work following the Sprint. The airport webcam project, which would typically have taken months, shipped two weeks ahead of its already tight deadline.
For Michael, the outcome was what he'd been trying to achieve for some time: an experienced team now working with AI as part of their standard process, moving faster without any change to the quality of work they'd built their reputation on.
Ready to build your own lean systems?
In two hours, we'll map where AI creates real leverage in your business and leave with a roadmap specific enough to act on immediately.
Book your Strategy SprintQuestions? Get in touch at hello@samanthanorth.com