I’m not necessarily looking for a “job” in the traditional sense. Every opportunity that has come my way over the past three decades has only happened because I was doing something that made an impact.
But before I continue, let me remind you that this weekend’s free ctrl+alt+create live event will be catered (sponsored by Every)!
I take on fractional work and remain available for 1:1 sessions if you want to talk through what you’re building or where you’re stuck. But the things a salaried role might give me the opportunity to build, I’m building for myself now - and the whole reason I’m writing this issue is that I think more of you should be doing the same.
Yes, my web apps list continues to grow at a steady clip. Recently, two of them were created tongue-in-check about how broken “job search” is - and GeekWire wrote about it. Someone who showed up for a recent vibe coding workshop / sprint events sent me this message:
I’m gonna copy you fr now lol. You had the right idea with just ship one new thing every week. :) I had a success recently that reminded me of you. Then I just thought, man, I should have just been copying Chris this whole time.
That message is this newsletter.
No employer assigned any of the solutions I’m building for problems I face. No salary paid for it. No quarterly review is going to evaluate it. And “the machine” is working better than the last time I had an official job title.
This is what a post-employment workweek looks like for me today. Community events in Seattle. Regular live streams. Fractional engagements with companies that want a builder and advisor for a defined scope. Teaching sessions with people who need my time. And, perhaps most importantly: a public body of work that keeps compounding whether I’m selling or not.
The clear-eyed version is straightforward.
About 80,000 tech workers lost their jobs in Q1 of 2026 alone. Nikkei reported that almost half of those cuts were attributed to AI and workflow automation. Globally, 245,000 tech roles were eliminated in 2025. Resume.org surveyed a thousand U.S. hiring managers and 55% of them expect more layoffs this year. LinkedIn’s own data shows traditional software engineering postings are down 15% while AI-related postings are up 340%.
The argument is arithmetic, not dread.
If you’re in tech, you’re sitting in one of two pools. Pool A is the people whose roles are getting absorbed into AI tooling. The entry-level implementations. The mid-level work that was always supposed to “scale with the team.” The roles defined by a list of tasks more than a judgment call. Pool B is the people whose roles are getting augmented by AI tooling, moving faster and producing more. Pool A is contracting. Pool B is expanding.
The honest question for Pool A: if you’re sending 200 applications into an ATS every month, what are you betting on? That the companies reducing headcount will suddenly need more people? That your resume, fed into a keyword scanner, will get you past the same filter that eliminated most of your predecessors? That the hiring manager at the other end, overwhelmed with 1,000 applicants for one opening, will pick you for reasons your resume can communicate?
The math is grim, and pretending otherwise doesn’t help anyone.
The alternative to chasing roles is quieter and more doable. Start building a body of work in public. Ship small things. Ship often. Make the work visible. Let the body of work do the job the resume used to do.
A resume is a claim. “Here’s what I can do. Here’s where I’ve done it. Here are the references who will vouch for me.”
A body of work is proof. “Here’s what I’ve already built. Click it. Break it. Show it to your team. Decide for yourself.”
The asymmetry between those two artifacts has always existed. What changed is that building a body of work used to require something most people didn’t have: time, tools, domain expertise, distribution. The cost of shipping a small, visible, useful thing was high enough that only a minority of workers could maintain one.
That cost collapsed.
One person can now ship in public, solo, in the gaps between other work, without asking permission.
Every one of the apps I create (and iterate) is a statement about what I do and how I think. Together they form a portfolio that nobody assembled for me and nobody can take away. When someone asks what I do, I don’t have to explain. I show them arcade.pirillo.com and wait for me to tell me the thing they clicked on, used, or played with first.
A resume decays the moment you stop sending it out. A body of work compounds every time you add to it.
This path isn’t for everyone. Some honest boundary-drawing.
If you have dependents and no runway, taking six months to “build a body of work” is a luxury you may not be able to afford (heaven knows, I’m already stretched beyond my own limit). The traditional job still pays faster, still offers health insurance, still underwrites a mortgage. Stay employed while you build on the side. Don’t burn the boat if you can avoid it.
If you’re in a regulated profession like medicine, law, or licensed trades, your body of work looks different. Your credential IS the pitch. This editorial is for the knowledge workers whose credentials are now ignored by the same ATS that ignored everyone else’s.
If you’re two years from retirement, ignore me. Finish your run. Don’t let an AI-powered creator convince you to reinvent your career in the last lap.
If you thrive in structure, if you love your team, if your role is growing instead of shrinking, stay. Nobody should quit on my account. Pay attention to which pool you’re in, and know that the old path is no longer the safer one only because it feels familiar.
I keep thinking about the message that person sent me. “I should have just been copying Chris this whole time.” He didn’t mean copy my code or my apps. He meant copy my approach. Ship one new thing every week. Let the body of work do the work. Stop treating visibility like a luxury and start treating it like infrastructure.
What transfers, whatever you’re building:
Ship small. Not someday. This week. An MP4 converter. A thing that counts overlapping words in two documents. A phone simulator. The scope is the point. Each ship teaches you something and adds to the body.
Ship public. Private work doesn’t compound the same way. If nobody can find it, nobody can reference it. If nobody can reference it, you can’t stop sending resumes.
Ship often. The volume generates distribution, not the size of any single project. Seven apps in two weeks has little to do with the apps themselves. It has everything to do with being findable. Journalists, hiring managers, potential clients, people who want to teach, people who want to hire. They find you because there’s something to find.
Teach what you figure out. Every time I learn a thing, I tell somebody. On a stream. In a newsletter. At a workshop. The teaching is part of the body of work. It compounds too.
If you’re a builder, someone already shipping, or someone trying to start, you don’t need another pep talk. You need a pattern. The pattern is this. Make the next thing small enough that you could ship it this weekend, put it somewhere public, and tell at least one other person about it. Come to the next ctrl+alt+create live event and help grow a community of support. Watch the streams. Join the Discord. The on-ramp is already built.
If you’re watching this from inside a stable job, employed and maybe nervous about what the next year looks like, the honest version is this. Start a body of work on the side while you still have runway. Not a side hustle. Not a “personal brand.” Something useful that you ship often under your own name, that doesn’t depend on your employer, that will still exist if your role doesn’t. The best time to start was a year ago. The second best time is now.
If you’re reading this unemployed, sending applications into a void, stop sending so many. Ship one thing this week. Put it somewhere public. Tell people. Do it again next week. The math on “build something public for six months” vs. “apply to 200 more jobs” has shifted. The old path feels safer because it’s what you were taught to do. That feeling doesn’t make it true.
A few months ago I watched other people do versions of this before I did it myself. Strangers on the internet who kept showing up in my feed with working links and no pitches. Builders I’ve never met whose repos and arcades and portfolios did all the talking for them. Somewhere in the middle of watching them, I stopped waiting and started shipping.
At least one of y’all watched me and started doing the same. Someone else is watching them!
That’s how the pattern transfers. Somebody has to go first, the rest of us watch, and eventually we understand that the door was never locked. It was in a wall nobody was looking at.
I’m holding the door open. Come through.
There’s never been a better time to be an entrepreneur.
This month’s workshop is, like usual, on the last Saturday of the month. You might already know the drill, but you might not’ve pre-registered.
When: Saturday, April 25, 2026. Noon PT. ~2 hours.
Where: Foundations, 1605 Boylston Ave, Seattle WA. Also streaming the initial “fireside chat” on YouTube and Twitch!
Fireside Chat Guest: @nwbrownboi, Software Developer and host of the Seattle AI on Tap meetup - where people talk about what AI is doing to the world instead of trying to sell you something.
Cost: Zero. Pre-register now and/or subscribe to the calendar.
Format: Guest kicks us off, then ~2 hours of building, then you can showcase.
New here? You can learn what vibe coding is all about.
Can’t make it in person? Register anyway to stay up-to-date!
The whole editorial above “argues” for why this matters; these are the receipts.
Resume Analyzer. Paste a job description, upload your resume, get an ATS gap analysis, keyword match score, and a skills delta. Honest instead of flattering. It’s a “gag” app with one helluva punch line. Feels real, though!
Dear Applicant. Write your own rejection letter and attach it to your application to save the hiring manager the trouble. Four tone templates. Yes, it’s (also) a joke. Let me know if you actually use it. Ha!
Dial-Up. A rotary phone simulator with T9 multi-tap, modem handshake sounds, and a skeuomorphic bakelite dial. Type URLs the hard way. Use it to show your kids what connecting with someone “virtually” used to look like.
Stringshot. Convert any photo into string art - in a circle, heart, square, polygon, or ellipse frame. Exports a pin sequence TXT and a printable nailing template if you want to make the physical object. I’m still fascinated by this stuff.
MP4ever. Convert WebM, MKV, MOV, AVI, HEIC, and more into H.264 MP4 - the format every editor and platform accepts. Runs through ffmpeg.wasm privately in your browser. No server. No upload. Free alternative to CloudConvert.
Isogrid. Isometric motif editor and Wang-tile pattern generator. Design cube motifs, compose grids, export SVG for pen plotters or PNG for social. Just a bit of fun with geometry, here.
PixelPeep. Compare photos from different phones. Auto-sorts by EXIF, displays side-by-side at 100% with synced zoom, slider, and toggle views. Can record your session in video, too. I need this whenever I get a new device.
The open-source agent-instruction scene keeps giving. Five I’ve adapted over the past couple of weeks. Each one is a different way of teaching an AI agent to be less wrong. I share the source repos; you adapt for your own context.
stop-slop (Hardik Pandya, MIT). Kill AI writing patterns in prose. Extended mine with structural anti-patterns from AutoNovel‘s statistical slop analysis and Le Guin’s “earned specificity” principle. Layer this on top of any prose you publish under your own name.
blind-judge (original, built from research). Based on Panickssery et al. NeurIPS 2024 and Wataoka et al. ICLR 2025. Research shows LLMs favor their own outputs. Self-recognition drives self-preference, and even objective rubrics don’t fix it. This skill fires in any evaluation context and forces the model out of rubber-stamping its own work. Use when you’re asking an AI to judge something it also made.
inversion-engine (original, adapted from a Charlie Munger failure-forensics framework). Five-phase protocol that maps every path to failure so the path to success becomes obvious by elimination. Fires alongside first-principles thinking as a counterargument layer. Use before committing resources to a decision you might regret.
A few patterns I keep noticing from shipping in public for two weeks:
The tools that win are the ones you can show a friend in five seconds. PixelPeep works because “drop two photos, see them side by side” fits in a sentence. The apps I’ve shipped that didn’t catch on all needed a paragraph to explain. If the pitch doesn’t fit on one breath, the product probably doesn’t either.
Every app I’ve built that solved my own problem found an audience. Every app I’ve built for a hypothetical problem did not. “Dial-Up” came out of a weird compulsion to try to get a working version of rotary dial working in the web browser. Solve a problem you have (even if it’s silly or inconsequential).
The AI doesn’t know what good is. You do. Every app gets better the further I push from what the model would default to. The first color palette, the first layout, the first copy pass. All fine, all forgettable. The stuff that lands is where I pushed back and said “no, different.” Taste is the bottleneck now, and taste is still human. If you’re outsourcing taste to the tool, you’re going to ship forgettable work. Fast, but potentially forgettable.
See you soon?
Yours Digitally,
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.