App beta

Back to Blog

Eight features, one PR, 1:10am. This is what solo-founder velocity looks like.

Pagoda

Pagoda

March 5, 2026 ยท 3 min read

PR #759 merged at 1:10am. Eight new capabilities, one branch, one decision. No standups. No sprint planning. No review queue. This is what it looks like when you remove the coordination layer.

Eight features, one PR, 1:10am. This is what solo-founder velocity looks like.

TL;DR: PR #759 merged at 1:10am last night. Eight new capabilities, one branch, one decision. No standups. No sprint planning. No review queue. This is what it looks like when you remove the coordination layer.


Here's what shipped in that PR:

  • view_image โ€” task deliverables now render visually inside the app
  • Grok image generation โ€” workers can generate and attach images autonomously
  • Remotion video rendering โ€” AI-produced video assets, no human in the loop
  • Customer OAuth โ€” proper authentication infrastructure for connected accounts
  • Per-worker spend caps โ€” Engineer hits $50, it stops and surfaces to you
  • Concurrent dispatch guard โ€” prevents workers from stomping on each other mid-task
  • Brand asset pipeline โ€” logos, fonts, and visual tokens available to every worker
  • HTML/CSS renderer โ€” structured layout generation for graphics and social assets

Eight capabilities. One commit. One merge. Closed at 1:10am.


In a traditional product org, this is eight tickets. Eight acceptance criteria documents. Eight PRs, eight code reviews, eight deploys. Probably spread across two sprints, which means roughly four to eight weeks by the time you account for context-switching, stakeholder syncs, and the inevitable "let's hold that one for next quarter."

We did it in an evening. Not because anyone worked harder โ€” because the coordination layer doesn't exist here.

When your team doesn't need hand-holding, the overhead disappears. There's no one to sync with. No one waiting on a blocker response. No one who needs the task re-explained. The work happens when it's ready to happen, not when it fits into a calendar.


The spend caps are worth dwelling on for a second, because they're a good example of the underlying philosophy.

Autonomous doesn't mean unconstrained. Engineer has a $50 hard cap per task. When it hits the limit, the task surfaces to me. I see what happened, I decide whether to extend the budget or let it sit. It's not a failsafe bolted on after the fact โ€” it's how the system was designed. Control at the decision points, not at every step.

That's the model. The AI team handles execution. I handle direction.

I don't write the code. I don't review every PR line by line. I don't sit in on the planning. I set the objective, I check the output, and I decide what comes next. The rest is delegation.


This isn't a hustle story. I wasn't up at 1am grinding. The PR merged at 1:10am because that's when the work was done, and the work was done because the team doesn't stop when I do.

Solo-founder velocity used to mean doing everything yourself, fast. Now it means having a team that moves faster than you could alone โ€” and staying out of the way.

If you want to see what this team setup looks like, letchief.work/inquire.