Disinto
from vision to market fit
Set the vision. Define quality gates.
Agents derive the backlog, build, review, ship, and operate.
You're the founder, not the developer.
A mining robot, lost and confused, builds a Disinto from scrap — a device so powerful it vaporizes three-quarters of a mountain on a single battery. Nobody can figure out how it works. When ordered to explain, it calmly dismantles the machine into rubble and flinders. — Isaac Asimov, "Robot AL-76 Goes Astray" (1942)
Write the vision, not the code
Describe what you want built. Agents figure out what's missing,
break it into tasks, and start working. You wake up to
progress, not process.
Nothing ships without your gates
You define the boundaries — what gets auto-approved and what
needs your eyes. Agents work within them.
You steer. They row.
Decisions, not busywork
The work organizes itself. Duplicates get closed, priorities
get surfaced, gaps get found. Your job is
direction — not tickets, not triage, not grooming.
The loop
vision
→
backlog
→
build
→
review
→
ship
→
operate
→
learn
→
vision
219
commits in 6 days
50
issues closed
2
human repos
Disinto building itself — and a DeFi protocol — on a single 8GB VPS.
One VPS, one repo, one battery
Bash scripts and Claude. No Kubernetes, no microservices, no SaaS dependencies. Runs on an 8GB VPS. Point it at a Codeberg repo with a Woodpecker CI pipeline and it starts building.
Built for web apps and crypto projects. SaaS dashboards, DeFi protocols, token launches — if it has a repo and tests, Disinto can work it.
Opinionated
Good defaults, few knobs. Works out of the box.
Bash over frameworks
If it can be a shell script, it is.
Pull over push
Agents pull work when ready. No orchestrator daemon.
Self-improving
Every solved problem becomes a pattern for next time.