E5 — Solo loop
What it is: Bansat’s own light loop for one person building alone, often with an AI agent. Team ceremonies make no sense on your own, but the discipline still does: plan a little, build in small slices, check the gates, and look back.
Fits: solo founders, students, researchers and freelancers working alone. Doesn’t fit: client work with formal stage approvals; put the solo loop inside E3.
The loop:
- Weekly plan (30 minutes): pick one to three FRs that meet the Definition of Ready (11.2), and write one sentence on what “done this week” means.
- Build: one item at a time (a WIP limit of one), in vertical slices: schema, server, UI, tests (12.2). Commit and open small PRs, even to yourself. For a stream of requests or a retrofit, keep the items on a simple Kanban board (E2).
- Weekly review (30 minutes):
- Demo what works to yourself, or better, to a likely user.
- Update the blueprint and the gap register.
- Note one thing to change next week.
- Monthly check: which gate are you working towards, and what’s left for it? Re-run the 17 launch traps (9.26) before G2 and G3.
Where Bansat fits:
| Bansat | In the solo loop |
|---|---|
| G0 | An idea note: Part 0 and 1.1–1.3, however rough |
| G1 | You sign the G1 row before writing production code. Signing your own gate is fine; skipping it isn’t. |
| G2 | A checklist before the first real user, waitlists included |
| G3 | Part 13 green on launch day |
| G4 | The 30-day review |
| High-tier projects | Get one outside review before G3 (How to use, depth by tier) |
Agent notes: keep to the week’s items; if a new idea comes up, add it to the backlog instead of starting it; at the weekly review, say plainly what isn’t finished.
How it commonly fails: endless polishing with no users; building without a blueprint because “it’s only me”; and skipping the privacy and launch checks because nobody is watching.
Measures: items finished per week (throughput); days since a real user last saw the product; open gap-register rows by severity.