Writing on product leadership, software, and the systems we build with AI.
I'm Rob Alford, a Melbourne-based product leader. I write about how AI is changing the way teams build software: context engineering, decision quality, and turning generic AI into a genuine product partner.
Latest writing
-
Diverge. Test. Repeat.
Converge Last: How to Use Design Sprints to Actually Drive Innovation
A decade ago I helped run a design sprint that broke most of the rules in the playbook: testing days early, with multiple ideas, and no executive super-votes. That's exactly why it worked. Here's the field guide for product organisations that want to manufacture innovation instead of waiting for it.
13 min read
-
Decide well. Revisit honestly.
Seeing It Written Down: On Being Profiled, and the Principles I'd Defend
Leaders Loop just published an editorial profile of me, and reading your own life narrated back to you is a strange, clarifying thing. It named a throughline I've spent twenty years half-saying out loud. So here are the four principles I'd actually stand behind, in my own words.
-
Diverge. Capture. Converge.
Think First, Write Last: How I Actually Write This Blog
I almost never start by asking AI to write the article. I start by using it to think, opening an idea up until something sharp appears, then arguing it back down into something I'd put my name to. Here's the actual loop, with real examples from conversations that became posts, specs, and a whole website.
-
AI is not the fad. Unbounded AI is.
The AI Overuse Problem: We Mistook a Brilliant Guessing Machine for a Deterministic Engine
LLMs are not a fad. But using them for everything might be. A look at where probabilistic AI creates leverage, where deterministic systems create trust, and why the future belongs to people who know the difference.
-
Memory. Coordination. Context.
The Innovation Elephant: Can AI Make Us Better Product Leaders, Not Just Faster?
Generic AI gives plausible product advice, and plausible is not the same as useful. This is the story of an experiment, turning a confirmation-bias chatbot into a genuine product partner by fixing the one thing that actually mattered: connected context.
What I write about
-
AI & decision quality
Using AI to think better, not just produce more, faster.
-
Context engineering
Why useful AI depends on structured, persistent, trusted context.
-
Probabilistic + deterministic
Combining LLM judgement with systems you can actually trust.
-
Product leadership
Strategy, decisions, and how teams actually build software.
These are working notes, not verdicts: true as I see them at the time of writing, and certain to age.
I hold strong opinions, loosely. Publishing them is how I pressure-test my thinking in the open: to start arguments worth having, and to be corrected by people smarter than me so we all learn something.
I'm already looking forward to reading these back and finding exactly where I was wrong, and watching how the thinking moved on.