🤖 How do I build Swingback alone with AI agents?

I started Swingback about 2 months ago, and most of the code is written with AI agents (mostly Claude Code). But it only works because of 3 things:
A plan before any code. Every part of the product is written down first and split in phases, and the agent stops at the end of each phase so I can review it.
One file of rules the agent reads before it starts. For example: every query has to filter by workspace, and internal notes must never reach your users.
Tests that check those rules. One test goes through everything your users' portal can show and fails if an internal note appears. Another one fails if the AI features can ever send an email or touch GitHub. So even if the agent forgets a rule, the tests catch it.
What I still do myself: decide what to build, review every phase, deploy and merge.
Are you building with AI agents too? What does your setup look like?

p/runevr
p/producthunt
p/google-cloud-platform
p/htmldrag
p/kodlo
p/ai-context-flow
p/dex-by-exmergo
p/jot-2-0
p/beamfy
p/refs
p/reso
p/duckfightclub
p/seshka
p/nudgeforme
p/talanoa-2
p/dino-kids
p/nexupira
p/many-displays
p/murror
p/supabase
Replies
Be the first to reply
Have a question or a thought to share? Add a comment above to start the conversation.