🌍 From “too many ideas” to working solutions — with AI as my partner
I’ve always had more ideas than I could possibly execute:
- Some could help communities
- Some could help businesses
- Some could simply make life easier
In the past, many stayed on paper — hiring a team of 45 developers and business experts wasn’t realistic (or so a younger version of me thought).
Today, I work with AI agents. They’re not a replacement for great human experts — but when used well, with a human in the loop, they can create good enough value fast.
AI gives me the leverage to turn ideas into working tools without the huge upfront cost. And because I believe in the power of open collaboration (a value I embraced as an OpenUK Ambassador), I always make at least part of my tools open source.
💡 My approach:
- Build tools for my own use first
- Share them — always at least partly open source
- Monetise where extra functionality or dedicated support is needed
🛠 The latest example — GitHub Cleanup Tool
I recently faced a messy, time-wasting problem: 325 GitHub repositories full of old projects, abandoned experiments, and duplicates.
Manually cleaning it up? Not going to happen. It just takes too long and would have helped only one person - me.
So in just a few hours — using AI to help me design and refine the build — I created a Python-based automation tool that:
- Audits all repositories and gives actionable summaries
- Lets me archive, privatise, or delete with interactive safeguards
- Protects critical repos automatically
- Exports results for reporting and compliance
This wasn’t about saving a few clicks.
It was about reducing mental clutter so I can focus on high-impact work — the same principle I apply to business processes when I look for automation and AI opportunities.
Interested? Here is more:
📦 Open Source & Full Write-Up Blog post with screenshots and stuff
📦 GitHub repo github-repositories-cleanup
💬 Let’s collaborate
If you have a business case that could be solved with automation or AI — and it fits within my employer’s (Mastercard) policies — I’d be happy to explore building it.
AI Assistance:
This article was drafted with the help of ChatGPT and then reviewed, edited, and fact-checked by me before publishing.
Top comments (0)