On The State of Software

You described SaaS as a "weird feudalistic regime" where people have been homeless on the internet. What does it actually look like to give people a real home in the cloud and why hasn't it happened until now?

Our experience of the internet is like being on a tour bus. We peer at the internet through windows: apps & websites built by other people. Sometimes they let us out to walk around, but our data and attention are ruthlessly monetized.

To truly live on the internet, you must have a permanent home, where you're allowed to construct anything you want, and interact with other individuals & businesses without middlemen.

Before, this was difficult to do on your own. Only technical, motivated folks could operate their own servers. AI is changing this dynamic, but AI alone is not enough. We must reimagine the cloud for the everyday person – and the personal agents that will help us navigate the new internet to come.

The personal computer was supposed to be the great equalizer. In what ways did that promise fail and how is the personal cloud computer a second attempt at the same idea?

I like this quote by Steven Levy (1984): "Access to computers—and anything which might teach you something about the way the world works—should be unlimited and total." This was the spirit of early personal computing. Zo is very much about bringing back that spirit.

Most people in the world have been left behind, when it comes to leveraging the full power of computers. Today, cloud computing is undoubtedly the most evolved form of computing, but only developers and tech-enabled companies are able to tap its potential. We must teach everyone how to fish, and give them access to better tools.

You mentioned that AI creates a Wozniak for everyone. Say more about that. What does it mean for someone to have their own Woz now, and what does that unlock that wasn't possible before?

Wozniak is the platonic technical cofounder: brilliant, sweet, and selflessly generous. He designed the Apple I and II almost single-handedly, and then gave away chunks of his own stock to early employees who got cut out. Every non-technical founder's fantasy is to find a Woz. Now, for the first time in history, anyone with an idea can summon a Woz: a technical partner who is endlessly patient, works all night, doesn't want credit, doesn't want equity, and can turn your half-formed sketch into a working thing by morning. When we think about crafting the personality and spirit of Zo, it's Steve Wozniak.

Zo's most enthusiastic users aren't always engineers: they're chiefs of staff, startup cofounders, and creative leads. Did that surprise you? What does it tell you about where the real demand for this is?

It's not that surprising if you look at the data – 50% of US adults don't use AI at all. There are so many non-technical people in the world who are just totally not exposed to the AI bubble. When we launched Zo, the pitch was "AWS for your mom". My mom was one of our first beta users, and quickly went from barely using AI to supremely AI-pilled. My mom has a server, and she uses Zo every day in her work and life. She's setting up automations to find cheap business-class flights to China, doing deep research on our neighborhood's housing prices to negotiate with real estate agents, and vibe coding data analysis & literature review tools to help with her biology research. It's still mind-boggling to me that my mom does all of this, all on a server of her own.

There's a growing pile of AI tools that are individually impressive and collectively a mess. Where does Zo sit in that stack and why is integration the hard problem that nobody else is really solving?

My cofounder Rob and I are proud digital hoarders. We've collected notes, files, side projects, datasets, and other digital exhaust over the years. When LLMs started getting good in late 2024, it was clearly a watershed moment for all this stuff we'd collected – finally, we had the leverage we needed to start doing cool stuff on top of it. We started building Zo out of personal necessity. We wanted a trustworthy, fully loaded, infinitely flexible home for our stuff. A place to work on everything with AI, and host anything on top of it all.

Because Zo is the home for the stuff you want your AI to access, it naturally becomes a hub you can connect to other AI tools. From the very beginning, we've designed Zo to be accessible programmatically (using our API or MCP), which makes it easy to use Zo as your context layer for any tool. For example, I connect to Zo from Claude Code to access my research & projects. I also often use Claude Code inside of Zo, to drive my Zo agent.

What does a good Zo setup actually look like for a creative or non-technical person? Walk us through a real workflow: what's running, what's connected, what happens while you're asleep or not actively using it.

I like to share an example from my personal site, 0.zo.space. If you scroll to the bottom you'll see it lists all the tracks I've been listening to. Every minute or so, Zo checks what I'm listening to on Spotify, finds the track on YouTube (since not everyone uses Spotify), and writes the track and YouTube URL to a little database on my Zo. That little database is so useful to me. It's part of my website. I'm using it for an automation that sends me a digest of upcoming shows. I can ask questions about my listening patterns. And I made all of this on a whim one Sunday, simply by texting my Zo while I was out and about.

If Zo works exactly as intended in five years — what has changed about how people relate to technology and to each other?

I like Venkatesh Rao's idea that AI is a ludic technology. Humans are playful creatures. The personal computer evolved out of play. Mainframes were huge and expensive – only accessible to large, tech-enabled companies. But a ragtag band of hackers & tinkerers formed communities like the Homebrew Computer Club. When I go to personal agent events like ClawCon, the energy in the room is electric. Everyone is so excited about tinkering and discovering what the future holds. And I wonder whether the Homebrew Computer Club felt like this.

The mainframe of today is everything inside the software company. Code, compute, collective knowledge, and agents that can operate inside that system, and expose new interfaces to the outside world. When we enable everyone on earth to have all of that, for their personal use, we will discover an incredible untapped vein of human creativity and potential. The mission of the Zo Computer Company is to enable this future. I can't wait to see it unfold.

Anything else you want to share? Is there a question we should have asked?

It's surprisingly hard to "turn a folder into a website". Say you have some files, like notes or photos, somewhere like iCloud or Dropbox. Why is it so hard to create a dynamic website based on your digital stuff in the cloud? Why do we have to do all this busywork to maintain a public presence on the internet? This question has gnawed at me for years, and the root cause is simple: we're just tourists on the internet. It's time to change that.

What are you listening to right now?

t0ni – Ego Death Memories. Literally listening to this right now, it's one of my favorite albums.

Last thing you read that genuinely changed how you were thinking about something.

This little blog post, Agents are actors, reminded me of a deep rabbit hole I went down in 2023 absorbing Michael Levin's research. My parents are both biologists, and I almost became one too – before I discovered how much I enjoy building software. In biology there's a spectrum of "agency": from single cells to organs, and organisms. In 2026, as the world begins to orchestrate digital agents, I think there's a lot to learn from biology.

Favorite tool, library, or product that most people haven’t heard of.

I'm late to the party, but I've been really enjoying Arc Browser. I'm anxious about their competing priorities with Dia (and Atlassian), so who knows how long my fun will last. Arc might soon become legacy software, but it sure is a fine vintage – browsing is fun again!