Why I Built A Million Pixels
I recently launched A Million Pixels.
I posted this on the A Million Pixels site, but including here for broader reach.
A Million Pixels just launched.
It contains a 1000×1000 pixel canvas, where you can select a block, generate an image with AI, and collaboratively build the canvas. That part is obvious from the homepage.
You might reasonably ask: why did I build this?
We’ve Crossed an Inflection Point
I'm a very curious person. As generative AI and LLMs have progressed, I've been increasingly adopting them. Recently, I have been using coding agents a lot. After using Opus 4.5 and codex 5.3, like many, I am convinced we've reached an inflection point.
The current generation of tools are not just powerful. They enable workflows and outcomes that were not previously possible. You can now build in hours or days what would have taken weeks or months. You can iterate at a fundamentally different pace. You can explore architectural alternatives conversationally.
It’s genuinely awe-inspiring.
And if you build software for a living, it’s at least a little unsettling.
Context
For those that don't know me, I've spent the better part of the last two decades helping technical people learn from each other and using a growth mindset to help companies earn durable trust with engineers. At least that's what my LinkedIn profile says.
What does that actually mean? I founded LinuxQuestions, I'm on the board of the CNCF and OpenSSF, and I'm the VP of technical community and open source at Datadog. At Datadog, I lead an org of about a dozen teams that operate at the intersection of engineering skills and communication skills. Some, like Technical Content, work mostly in natural language. Some, like the team that maintains Vector, work mostly in code. Some, like Customer Education and Advocacy, are in the middle. Those experiences mean I've seen the impact of AI across a wide variety of skills, vocations, and organizations.
And I have opinions.
What this is... and isn't
I don't think the reality has set in for many people yet.
On one side, you have claims that SaaS is dead and software engineering is going away.
It isn’t.
On the flips side are the thoughts (feelings?) that AI is a fad and is going to disappear or fade away.
It won't.
While you may not be able to blindly trust AI output (yet?), we're way past the point where you can just ignore it. Software developers aren't going away, but there are very real implications for anyone who builds software.
The Internet changed everything. I believe this will too. I also think there are a lot of similarities between the Internet-era bubble and where we are today, but that's a topic for a future post.
The Next Abstraction Layer
It's easy to forget how adaptable software engineers have had to be over the years. In a lot of ways, coding with codex or claude code feels like the next level of abstraction to me. We went from punch cards, to very low level languages, to higher level languages, to this. Each layer allowed us to focus more on outcomes and less on the implementation details of the language.
But here’s the part that gets lost in both the hype and the backlash: abstraction doesn’t eliminate fundamentals. It makes them more important.
At least for now, you very much need to understand foundational engineering principles to use agentic coding for anything more than toy apps or cool demos. You cannot build something durable, scalable, or secure without deep engineering judgment.
Agentic coding amplifies skill... it doesn’t replace it.
Why I Built A Million Pixels
So back to the question posed in the first paragraph: Why did I build this site?
Two reasons, and I've already alluded to them both.
First: curiosity
I wanted to see what I could build. In the end, I learned a ton and the site was really fun to build. I'll cover a bunch of the details in future posts.
Second: conviction
I am convinced that the next generation of developers will need to be AI-native to be successful. As an industry, we are woefully underprepared for that future.
I don't think we really know exactly what AI-native means yet, but it's more than just the prompt. Knowing how to collaborate with systems like this. Decomposing problems well. Validating outputs critically. Leveraging agents responsibly. Who knows what else.
Access to that kind of experimentation shouldn’t be gated.
That's why 100% of the net proceeds (ie. after fees, usage charges, and taxes) will go to donations to coding education and providing AI credits. I am committed to not profiting a single dollar as a result of this site.
If this shift is as important as I believe it is, we need to lower barriers... not just talk about the future.
Build with me
So as the homepage says: claim a block, write a prompt, and watch AI (and humans!) create a unique canvas.
The About and FAQ pages likely answer many of the questions you have, but I'm curious to know what you think of the project. Use the contact form to let me know.
We’re all learning how to build in this new era. Let’s experiment in public.
--jeremy