The problem
When you build with agents, a feature rarely stays in one place. It can:
- start ina brainstorming session
- change aftera customer call
- split acrossagents and worktrees
- land in the product through one of them.
Exploring several directions is useful. The problem begins when those paths pile up. One worktree follows an older decision. Another contains an idea you already ruled out. The feedback that changed the plan lives somewhere else. Every piece exists, but the current direction is scattered across them and only you know how it fits together.
The faster you build, the more paths you create, and the harder it becomes to see which direction is current, why you chose it, and whether the work still follows it.
What Artifact is building
Artifact connects those paths into one history: customer conversations, decisions you accepted, agent sessions and worktrees, and the implementation that followed. It also keeps the recorded alternatives you explored and why you ruled them out.
If one path keeps following an older decision after new feedback changed the plan, Artifact surfaces the mismatch and shows the original sources on both sides. You decide what changes. Artifact makes the history easy to inspect.
Artifact separates what you explored from what you accepted. A rough idea or temporary instruction does not become product direction just because an agent acted on it. Getting that boundary right is a large part of what I am working on.
Who it is for
Artifact's beachhead is solo founders who build software with AI coding agents most days.
- talk to customers
- make product decisions
- run several agent sessions at once
The output looks like a team; the memory holding it together is still one person. They feel the direction problem first because no built-in partner acts as a second memory or constant mirror.
The larger vision is small AI-native teams. The same gap grows across more people, tools, and agents.
Momentum
I'm working on Artifact every day, with product and go-to-market moving together.
- ProductShipping, testing, and improving the product daily, turning what I learn into a tighter build.
- Go to marketTalking to founders, sharpening the beachhead, and mapping the real use cases.
I've spoken with more than twenty founders in the past three weeks. I'm obsessively trying to understand how they build, where direction breaks, and which use cases are worth solving first.
About me
I'm a designer at heart. I care about how a product feels, looks, behaves, and whether it solves the user's problem.
Product taste, product judgment, and the ability to turn both into something real. I can move from a rough idea to an interface to a working product without losing the reason it should exist.
I don't see design as a separate step. It shapes which problems I choose, what I remove, how the interface behaves, and where the quality bar sits. I want the products I build to feel clear, intentional, and alive in someone's hands.
For the past year and a half, I've been building nonstop with AI agents. I've tried many workflows and learned where these systems are strong, where they drift, and what it takes to get great work out of them. They give me enormous speed. My job is to bring the taste, context, and judgment that speed still needs.
That is why Artifact is personal. I know what it feels like to have several agents moving at once, ideas changing quickly, and the product living mostly in your head. I'm building the thing I need to move faster without losing the direction or lowering the bar.
I want to see how far a very small team with strong taste and the right tools can go.
Outside the work, I read philosophy and spend a lot of time with history. Both give me a longer view of people, decisions, and the ideas that survive.
Join Artifact
I want to build Artifact with someone who moves naturally between the system, the product, and the company.
The work is ambitious. I want to do it with someone who cares about the same things I do:
- Be ambitious, and make things happen. Do the work, put something real in front of people, learn from it, and keep moving.
- Stay very close to customers. Talk to them often and understand how they actually work. That is where product judgment comes from.
- Care about the whole thing. The brain, interface, product, story, and company all shape each other.
- Reduce scope, not quality. Ship less when needed, but be proud of what goes out.
The two hard problems in front of us
- The brain. Artifact has to build an accurate context graph from messy conversations, decisions, agent sessions, and code. It needs to know what was explored, what was actually decided, and how sure it is. If this is wrong, nothing else matters.
- The interface. All of that complexity has to become simple for the person using it. You should be able to see what changed, why it changed, where the evidence is, and what needs your attention without digging.
Let's work on a real problem and see how we think together.
Join the early group building Artifact with me.
For founders building software with Codex or Claude Code most days. You’ll get:
- Direct access to me. Ask questions, share where your workflow breaks, and help shape what I build next.
- Every product update. See the work in progress, what changed, and the thinking behind it.
- A small community of founders building with agents. Share tips, real use cases, and what we are learning. I’ll share what I learn from building Artifact this way too.