~/published
33 posts, newest first
── 2026 ──
I told my team I couldn't carry on as we were, and I had no idea what the alternative was
I told my team I couldn't carry on as we were, with no idea what the alternative was. Naming a problem you cannot solve converts a private worry into shared work, and the shared work is what produces the answer.
You cannot understand a team from the outside
I have never managed to assess a team properly without spending real time inside it first. The humans, the technical picture and how work moves all have to be read at once, and none of them is sufficient alone.
Rolling out a new tool takes a sprint
Rolling out a new tool takes a sprint. Getting people to change how they work takes months, if not a year, and waiting for people to be ready is the hardest part of leading change for me.
The most useful thing in a retrospective isn't the feedback itself
The most useful information in a retrospective is rarely the feedback itself, it's who speaks, who stays silent, and how people treat each other when something uncomfortable is on the table. Team health decays like technical debt, quietly, until you notice only once something breaks.
The hardest part of stepping up to lead wasn't the meetings
The hardest part of stepping up to lead wasn't the meetings or the politics, it was watching my team solve problems differently to how I would have, and keeping my hands off the keyboard. Micromanaging is distrust dressed up as standards.
My daughter was three when she told me she couldn't be an astronaut
My daughter was three when she told me she couldn't be an astronaut, not because of the maths or the danger, but because she wasn't a boy. The doors that close on people are rarely slammed; they close through a thousand subtle cues, until we finish the job ourselves.
You won't understand AI until you've used it yourself
There's an entire industry devoted to telling people what to think about AI, but reading about it isn't the same as understanding it. The fastest way to learn what these tools can do is to point one at something in your own week.
The best teams I've built were never the most talented on paper
The best teams I've built were never the most talented on paper, they gelled as a unit, could handle both a sprint and a marathon, and kept being reshaped as the people in them changed. Building a team isn't a one-shot decision, it's work that never really stops.
People assume agile teams skip the planning
People assume agile teams skip the planning. In my experience it's the opposite, I plan more as an agile lead than I ever did on a waterfall project. The difference is when the planning happens, not whether it does.
The pressure to do more never stops, it just changes shape
The pressure to do more never stops, it just changes shape, from 'one more task tonight' as a dev to 'do we stretch or rest' as a lead. The difference between catching it early and drawing the line too late isn't willpower, it's asking what's actually going on.
Every sprint the team's work is visible. The manager's isn't
Every sprint the team's work is visible, tracked, inspected and improved. The manager's work is nowhere to be found. I started putting my own commitments on the board, and it levelled accountability in a way no 'open door policy' ever could.
They didn't think they could build something like that
Someone I'm mentoring told me they didn't think they could build something like that. Two hours later they'd pushed a working app to GitHub, proof that the real bottleneck now is rarely capability, it's the decision to start.
I used to kill ideas before they had a chance
I used to kill ideas before they had a chance, the mental maths of cost and bandwidth would quietly bury them. AI has removed that filter, so the bottleneck has shifted from capacity to build towards the quality of what you choose to build.
The hardest lesson I learned as a tech lead wasn't about architecture or code quality
The hardest lesson I learned as a tech lead was knowing when to step back. Reviewing every pull request wasn't leadership, it was a bottleneck wearing a leadership badge.
The best teams I've led weren't close from team building exercises
The best teams I've led weren't close because of team building exercises, they were close because we'd earned the right to relax together. Building a team people want to be part of starts with the lead deciding relationships matter as much as the backlog.
Early in my career I was told not to use open source software
Early in my career I was told not to use open source software, then not to trust Stack Overflow, then not to use ChatGPT. Now it's AI coding agents, and the outcome is always the same: the tools win, so the only question is whether you adapt before or after it costs you.
How quickly can a new starter commit code?
How quickly can a new starter commit code? That question tells you more about your project's health than most metrics on your dashboard. A slow time to first commit usually points to fragile setup, missing tests, and outdated documentation papering over the gaps.
Your customers have already accepted problems you haven't found
Your customers have already accepted the problems you haven't found yet. I watched someone switch tabs and copy data manually every time they used a feature we'd built, a workaround they'd never think to mention because to them, it was simply how the system worked.
Every skill that transformed my career, I learned before I had to
Every skill that transformed my career had one thing in common, I chose to learn it before anyone told me to. Playing around with ChatGPT a couple of times isn't learning, it's tourism, and the gap between people actively learning AI and those waiting is accelerating.
Most companies run their AI strategy like a waterfall project
Most companies are running their AI strategy like a waterfall project: a committee, a strategy document, a phased rollout, while the technology shifts three times before phase one is done. AI rewards small experiments and fast feedback, not the perfect policy deck.
I used to get frustrated that designers couldn't just get it done
I used to get frustrated when designers couldn't just 'get it done' in a sprint, until I realised I was asking them to work against the very thing that made them valuable. Different minds aren't a coordination problem to solve, they're the reason the work is good.
I still believe in 20% time
I still believe in 20% time, not because it's trendy but because sprint-to-sprint thinking creates tunnel vision. The dividends don't show up this sprint, they show up in the team that's still sharp and engaged six months from now.
I spent two weeks building a filtering system nobody needed
I spent two weeks building a data filtering system with custom views, then watched a user export the table to Excel and combine it with everything else she was working with. Now 'show me how you actually do this' is where I start, not where I end up.
Technical debt doesn't announce itself
Technical debt doesn't announce itself, it builds quietly while you're shipping features and hitting deadlines, until a change that should take hours takes weeks. The speed you're celebrating today might be borrowed from next year.
How long would it take to build an app? I told him a weekend
The technical barrier is nearly gone: what used to take me three weeks now takes three days. When you can build almost anything, the bottleneck shifts to time, ideas and focus.
For 3 sprints in a row your team overdelivers, you face a choice
When a team overdelivers three sprints in a row, the instinct is to raise the target. I learned the difference between a team that's performing and a team that's sustainable, and gave the spare time back as an investment.
The hardest moment in my leadership journey wasn't a project failure
The hardest moment in my leadership journey was realising I was no longer the best engineer on my team. Your role as a lead isn't to have the best answers, it's to build a team that doesn't need you to.
Most advice you give disappears into the void
Most advice you give disappears into the void. Months after an AI workshop, someone sent me an app they'd actually built, a reminder that the feedback loop on encouragement is mostly invisible.
I've worked in teams of 3 and teams of 300
I've worked in teams of 3 and teams of 300. Neither is better, they're different games entirely, and the real advantage comes from experiencing both.
There's a line in the BlackBerry film that struck a nerve
'Why do you think these guys work 80-hour weeks?' Fear works short-term and so does money, but commitment comes from feeling heard, having autonomy, and being seen as a person. The stick gets compliance, the carrot gets discretionary effort.
── 2025 ──
I've read the same book every year for 10 years
I've read The Phoenix Project every year for 10 years. Not because I forgot what happens, but because I'm not the same person who picks it up each time.
I still use RSS feeds
I still use RSS feeds to keep a pulse on the trends that matter to me. What separates people in this industry is the habit of continuous learning, so always have a tiny project on the side.
I've spent years giving career advice in 1:1s
I've spent years giving career advice in 1:1s and never shared any of it publicly, because I was scared of what people would think. So I'm trying something: try it, learn from it, iterate.