I gave a new talk this week. The first one since last year. Not based on my book, unlike last year's - a truly, from scratch, 45 minutes on something I really knew nothing about a year ago.

Everything I learned about management that helped me build a - reliable - software factory. Basically.

Amazing how I've given so many talks and yet each one still terrifies me. It's getting worse, if anything. More pressure now, like I'm not just building a reputation, I'm maintaining it.

And also the validation. I think I've grieved the application formerly known as Twitter enough, but each time it hits anew. So weird to give a talk, have such gorgeous slides (Joe hit it out the park again), and then the outcome is... one LinkedIn mention.

Well, and the people who approached me in person. That is of course very meaningful. But what about the introverts? Same for me in the audience I guess, I had to catch the speaker I loved - truly the most unhinged conference talk I've ever seen - and tell her that directly rather than via @ mention.

This month, it feels like visibility is the job. This talk, another next week, two webinars - one Anthropic, the other Dev Interrupted. The continual round of social media, trying to remind people in funny, charming, less obvious ways, but ultimately - we built this thing, and you can buy it.

It's weird, how the thing I hated most in my last job and the thing I hate the most now are the same thing manifesting in different ways. Self promotion. In an organisation we call this "managing up" and "building visibility" but essentially it's the same - it's not enough to do the work; to get paid for it you have to convince people you did it, but not in a way that rubs them the wrong way.

And then in this portfolio career, the same thing. Although it feels a bit less of a pointless game, and a bit more fair. Of course, if I want people to buy a DRI course, I have to tell them about it. Of course, I have to make sure the funnels on the website convert effectively. It's not my favourite part, but it's all a piece of building this thing that I do love.

I am a regular reader of social media posts that are, in effect, "women's achievements hidden by the patriarchy" stories. They give me such mixed feelings. I love to know their stories. I hate that this is how I learn them.

And the corresponding ones, of "men who took credit for the work of others". Like the architect Frank Lloyd Wright. Around half the drawings in the portfolio that made his name in Europe were by Marion Mahony Griffin. He didn't credit her.

We all know that for women there's a line between being a bitch and being a pushover. These stories make me wonder if there's another line - one between getting credit for your achievements and getting sh*t done. I know I have walked that one. I was so effectively doing someone's strategy for them, until someone above realised. They stopped talking to me. I got some more credit, at the cost of actually getting what I needed done, and a side of resentment that felt undeserved.

I think it's a choice women make, consciously or unconsciously, and it's why so many of us hate the self promotion game - a choice between getting credit, and being resented.

I remind myself of this on the days it seems exhausting. That a change is as good as a rest, and self promotion does at least feel preferable to the corporate games that make up the alternative.

What I've been doing

Twill is two months past the replatform, and stable. We're out of the reactive post-launch phase and working on bigger things again, which is super fun. The new platform is now the baseline, not the goal.

I wrote two posts on how we run software factories. One on what managers even do now: ICs running agents are managers now, and need the pieces of management that are useful right now. Not how to run a 1:1, how to think one level up. The other on good architecture as a function of time: making good decisions easier, on a timeframe you can reason about, and why that is a human responsibility.

I'm so excited about Breaking Into the Security Mindset, our first partner course, with Halle Winkler. It's the kind of systematic thinking ICs need to work on right now. The first cohort starts on the 12th of October, and the intro module is free to preview.

The DIY MBA continues. The Coursera marketing course did not go well, so I read Obviously Awesome by April Dunford instead - the book I needed. Used the thinking from the book, the worksheets, and Fable to push through the positioning block and deliver new messaging all over the site. Next up: her other book.

The next DRI Your Career cohort starts on the 19th of October, our last of 2026. Wild that we are now in Q4! We have a few more exciting things in the works I hope to share before the end of the year.

Autumn speaking is under way. "then | now - velocity through rigour" at Bug Bash Europe in Copenhagen is done. Next week it's "DRI Your Career" at next.app devCon in Berlin, then the Anthropic webinar with Jean on the 13th of October, and the Dev Interrupted roundtable on the 29th.

On a personal note, last month, along with more than 5,000 other migrants, I became an Irish citizen. And, once again, a citizen of the European Union. Eight years in the making, Brexile is complete.

What I've been reading

• A Taxonomy of Tech Debt from Riot Games - Really helpful, and it gives me language for some of the disconnect between developers and people who think you can just AI everything. You need to be fanatical about the fundamentals (like data) because they impact everything. A mess there will slow you down later. And MacGyver debt: I've caught Claude doing this many times as it tries to be helpful. Non-engineers don't know what is and isn't MacGyvering, i.e. what will bite you later.
• The Structure of Engineering Revolutions by John Allsopp - on engineers resisting AI as a Kuhn-style paradigm shift. It is a lengthy - but compelling - argument on why it's necessary to accept the grief and do the learning. One thing I think is missing is that working with this effectively means moving one level up. It's similar to the IC to manager mindset shift - notoriously hard. But those changes are always easier chosen. For AI, engineers are having poorly articulated change forced on them. Of course they are finding it difficult. I think overall I'm more hopeful than Kuhn, because our view of the science is more expansive and has less god in it (cf. Einstein, "God does not play dice"). But maybe I'm just an overly optimistic atheist.
• The Compass and the Navigation from The Reframe - Set the direction, make the plan, expect the plan to change. I love the point on critique. Critique is only useful when it comes within a commitment to the direction - is it arguing about the plan, or whether you should start at all?
• The Sandboxed Local Agent by Halle Winkler - Very intense post on what it takes to really sandbox a local agent. The goal is (close to) zero trust: the agent can read and write in one working directory and nothing else. Getting there takes a hardened local model server, a sandbox profile with nearly everything switched off, a DNS workaround, and then 42 tests to check the boundaries are where you believe them to be.
• God in the box by Cory Doctorow - A good, calm explainer on the "AI taught itself to hack" story. Not novel, standard techniques - including those developed by middle schoolers - plus a lack of basic supervision. This is why my worries for humanity are driven more by WWIII than AI.