A publication from a 10-person team about how we actually build with AI. Not theory, not listicles — what we tried, what broke, what shipped.
I'm six years into building Attune. Any founder will tell you the journey is a roller coaster — the highs, the lows, the "are we going to make it" months, the "we just signed a huge client" months. But right now, I'm having more fun than I've had at any previous point. And I want to talk about why.
We kept having this experience: we'd read about a new AI technique or see a company get buzz for some novel workflow, and we'd realize we'd already built it internally — and then some.
We weren't trying to be cutting-edge. We thought our stack was pretty boring, actually: Claude Code, markdown files in git repos and Google Drive, some basic principles to make information routing and recall fast.
We built our AI systems piece by piece to solve our own business problems. A post-meeting workflow that logs every decision and action item automatically. An automation that reconciles credit card transactions against projections in minutes. A shared knowledge system that connects the dots as we work, so we make each other better. We did all of this so organically, as a side effect of just trying to run a small company well, that we didn't realize how far ahead we were.
We'd go looking for people at the forefront of using AI — people we could learn from — and we'd realize we were already there. Or past it.
So we started asking: Who are our peers? Where are people at a similar stage doing equally weird, ambitious stuff?
And we realized: if we want to find those people, we need to be visible. We need to share what we're doing.
/build is a publication from the team at Attune about how a small team actually builds with AI. Not theory. Not "10 prompts to transform your work." The real thing — what we tried, what broke, what shipped, and why any of it matters.
/build is named after a slash command — the shortcut you type in the terminal to get things done.
We're a 10-person financial health infrastructure company — built inside the Financial Health Network, funded by Incandescent, designed to be profitable really, really early. Not the usual startup story.
The articles come from the whole team. Everyone who ships something interesting writes about it.
I was hesitant to write about AI because I didn't like most of the noise.
I didn't want to write another listicle. I didn't want to cover breaking news about the latest model release.
Most of all, I didn't want to shame anyone. So many of the posts I see now are rooted in shame, guilt, and fear: If you're not doing this yet, you're falling behind. If you're not using this tool, your job is at risk.
Shame-based messaging is really not my jam. That's not how we operate internally.
Instead, we operate with dad jokes and laughs and a genuinely ridiculous amount of fun building things.
Every day, I am astounded anew at how fast it is to create something new. The distance between imagining something and actually seeing it live — clicking around, playing with it, testing hypotheses — has shrunk dramatically. Is it easier to create slop? Yes. Is it also easier to produce wildly different design directions, compare and contrast them against each other, and learn something new in the process? Also yes.
We can all prototype hi-fi at speed.
When I first showed Ksenia the terminal, she thought she'd never use it. She does finance and operations. She lives in spreadsheets.
Now she's the biggest champion, showing ops and go-to-market teammates how they can solve their own challenges in the terminal with Claude Code. She built a slash command to shrink credit card statement reconciliation from 90 minutes to 10. She built a skill called /dad-joke that gives you a very thoughtful dad joke when you run it. Both of those matter: the dad joke got the rest of the team comfortable with slash commands, and the reconciliation saves her real hours every month.
That path — from "I'm not an engineer" to building genuinely useful things — is available for every function.
If you've always had big dreams but didn't have the resources or didn't want the VC path, this might be for you. If you're a 5-person startup, a founder just getting started, or a cool social-impact organization doing interesting work — you might find something here.
The best way I can describe who this is for is emotionally. This is for people who have a nagging intuition that they can experience more joy in their work. Who want to get lost in the flow of making things. Who are tired of the hype cycle and the guilt and the "you're falling behind" energy, and just want to build stuff that matters to them, in alignment with their own values, for their own goals.
My background is in anthropology and science and technology policy. I've been studying how people and societies respond to technological change since long before I worked in tech. And I do think we're at a genuine inflection point — this is a before-and-after moment, like the internet, like the telephone. My response isn't panic or hype. It's curiosity. Asking: Who am I, what do I value, and how do I use these tools to build things that serve what matters most to me?
In other words: This stuff is weird, right? Here's what we learned. What are you learning?
We're trying to find our people. If something here resonates — if you're building something cool, or figuring out how to use these tools in your own quirky way, or you just want to be in conversation with other people who are having fun making things — subscribe, and tell us what you're working on.
Build logs, system designs, and honest tool reviews from a small team building with AI. No spam. Unsubscribe anytime.
Check your email — a confirmation link is on the way.
Something went wrong. Please try again.