No standup. No sprint planning. No ticket queue. Just you, a problem, and an AI pair programmer that never gets tired of your questions.
Building solo sounds like a fantasy until you're three hours deep in a bug that a teammate would have spotted in ten seconds. Then it sounds like a punishment. The reality is somewhere between those two — and it's shaped by how you structure the work.
What solo actually means
Solo doesn't mean isolated. It means every decision is yours — the product calls, the architecture, the design, the copy, the support emails. You are the entire stack: PM, designer, engineer, and customer service rep. AI tools fill in the gaps where a team would otherwise carry the load.
That's a lot. But it's also fast. No approvals. No consensus-building. If I want to change the entire design system of a product, I do it in an afternoon. If I decide a feature is wrong for the product, I cut it without a meeting.
How I structure my days
I don't follow a strict schedule, but I do follow a rough pattern:
- Morning: design and product decisions — when my head is clearest
- Midday: actual building, working alongside AI tools for the bulk of the implementation
- Evening: testing, emails, reading, planning the next day
- Never: meetings (there's nobody to meet with)
The hard parts
The obvious hard part is that some bugs take longer to track down. There's no one to rubber-duck with in person — though an AI assistant fills that role more often than you'd think. You still have to be your own final reviewer, your own QA, your own devil's advocate.
The less obvious hard part is momentum. With a team, someone else's energy can pull you forward. Solo, when you're stuck or uninspired, there's nothing external to push you. You have to generate that yourself every day.
What I've found that helps: keeping the scope of each work session small. Don't sit down to "work on the product." Sit down to fix one specific bug, or build one specific screen. Small targets are completable. Completed things feel good. Feeling good keeps you going.
Why I wouldn't change it
Shift Desk, SignaCard, and Audio Node exist because I had a problem and I could just… go build the solution, with AI accelerating the parts that used to take a whole team. No pitching the idea internally. No waiting for design resources. No roadmap negotiation. I had the itch, I scratched it, and shipped something real with users who find it useful.
That feedback loop — problem to product to people using it — is tight and direct and deeply satisfying in a way that's hard to replicate when there are ten people between the idea and the thing.
If you're thinking about going solo on a side project: do it. Start small, define the scope tightly, use every AI tool available to you, and ship something real. The first version doesn't need to be good. It needs to exist.