Last time I signed off with "catch you back here at the end of June." That did not happen, and neither did July. No interesting reason — the months filled up and this was the thing that slipped.
Which is a slightly awkward thing to write in a newsletter about operational discipline, so let me be straight about it: a monthly cadence we miss twice isn't monthly. We're keeping the monthly commitment, and if we drop it again you'll get an issue that says so rather than silence.
So this one is a catch-up. Two months, in one go.
Published in June and July
Eleven notes went out while the newsletter didn't. These are the four I'd start with.
Articles
Thinking
The most expensive bug is overconfidence
The failures that cost the most rarely come from technology. They come from people believing they understand a system more deeply than they actually do — and from organisations that never test the difference.
Read
Technology
The agent that runs your operation can't be on call
Agentic AI is being adopted faster than it's being trusted, and for good reason. An agent that takes actions in your operation isn't a faster tool — it's a new owner of decisions nobody can be paged for. The work of making it ownable is the same work it has always been.
Read
Thinking
Are you behind?
"If you're not using AI, you're behind" is the most effective sentence in business software right now, because it fuses something true with something manufactured. Pulling the two apart is the whole job — and it changes what "behind" even means.
Read
Thinking
How much does custom software cost in Australia?
A straight answer on custom software cost in Australia — real ranges, what actually drives the number, and why the quote you're handed before anyone understands your operation is usually the least reliable figure of all.
Read
Jayben's piece on overconfidence is the longest thing we've published and the one I've had the most conversations about since. If you only read one, read that. The other three are all versions of the same question from different angles: what you can actually know about a system before you commit to it.
Also published
Thinking
How might we: turning a vague worry into a problem worth solving
"How might we…" is a small language trick with an outsized effect: it turns a vague operational worry into a problem a team can actually build against. What the framework is, how to write good ones, the scope trap that ruins most of them, and how we use it before writing a line of software.
Read
Thinking
Getting off spreadsheets won't fix this
The case against your spreadsheets is usually the wrong case. The data structure problem existed before the first cell was filled in, and it will survive the replacement if you build to the same brief.
Read
Technology
Fable 5, in plain English
Anthropic's most powerful AI model was released, pulled by the US Government, and is back today. Here is what is real, what is not, and what a non-technical business owner should actually do this week — with sources.
Read
And two months of things that were less about theory and more about where we were.
Company updates
Company
One year in
A year ago Jayben and I started Slash Tech to fix one thing — good people being held back by technology that was never built for them. Here is what has held up, what has changed, and what the second year is about.
Read
Company
A week in Singapore, and what stuck with us
One of our founders, Javier Bates, was named a Westpac Young Technologist and spent a week in Singapore on its Disruptive Leaders Program. A short note on what the week reinforced — about people, leadership, and what's possible.
Read
Company
A weekend in Boulia, and what the field teaches
While one of us was in Singapore, our co-founder Jayben was in outback Queensland teching a musical performance for a friend's live-events business. A short note on what a weekend where nothing was guaranteed reinforced about the way we work.
Read
What we shipped
The headline is that two systems we'd been talking about internally for a while are now live and public.
Field Notes itself got rebuilt while we were at it. Every note now sits in one of three categories and carries a couple of tags, so you can filter for the thing you actually came for instead of scrolling — try Operations or AI and you'll see what I mean.
We also put our internal R&D work on the projects page for the first time. Those are the experiments we run on our own time and our own hardware — no client, no deadline, and the findings feed straight back into how we scope real work.
What's active
Coming up
That's June and July, finally. If something in here landed close to home — a system you're unsure about, a list you couldn't produce, a project you're weighing up — grab a Fit Call. Twenty-five minutes, free, and we'll tell you straight whether we can help.
Back at the end of August. This time on purpose.
— Javier