For the people who build, own, and operate.

InsanelyGreat

Enabling people striving to make their own dent in the Universe.

Reference Architecture · v0.9.5
Domain-Adaptive Manufacturing
Software that builds and configures itself to the reality of the manufacturer. From a two-person startup to Nestlé or Apple. Standardize the physics, bespoke the experience.
Read the architecture →
Methodology · v2.14 · SSD 2.0
Shippable States Development
Ship every day. No exceptions. A pragmatic AI-powered discipline that keeps your software deployable at all times — now one progressively-disclosed surface for newcomer and expert alike.
Learn more →
Manifesto
Agile2
The ability to be agile about your agile. Process is a tool, not an identity. If it doesn't increase velocity, change it.
Read the manifesto →
The After Effort Project · Course & Software
After Effort
The disciplined use of AI in teaching. A tool helps a student most when it comes after their own effort, not in place of it — the professional-development course, and its software companion, UpwardTeaching.
Explore the project →
New in v2.14
You can now let the agent run the loop. It still can’t ship.
The orchestrator has always proposed and waited. The autonomy ladder adds two opt-in rungs above that: advance, where bare /ssd executes its own unambiguous next step — one phase per invocation, never more — and /ssd run <slug>, which walks design → code → review → gate unattended, looping coder and reviewer until the gate passes or a budget stops it. The default is unchanged: with no autonomy block configured, the orchestrator makes no additional call at all.

What it cannot do is the actual design. ship, deploy, rollout-advance and flag-removal are refusals with no bypass flag — because a feature is bounded by two irreducibly human decisions, the brief at the start and the ship at the end. Autonomy delegates the middle and neither end. Every transition is announced, written to disk, and only then executed, so a run that dies mid-flight is recoverable rather than mysterious. And the run records the gate script’s exit status rather than its own opinion of how it went.

Also new: rail deviations are finally real. The rails promised since v1.15 that every skipped step gets recorded; measured across 15 workstreams, nothing had ever written one. A writer and a gate rule that reads it back now ship together — and an auto-run records zero of them, by construction.
See what's new

Essays — from teams of one to organizations of thousands

Opinionated articles on how work actually moves — at whatever scale you are working at.

At organizational scale

Telemetry

Instrumenting the Organization

Most companies measure activity. Flow is the thing that pays. How to instrument an organization the way you would instrument a factory — and what AI changes about the instrument.

Reference Architecture

Domain-Adaptive Manufacturing

Software that builds and configures itself to the reality of the manufacturer — from a two-person startup to Nestlé or Apple.

For solo developers and small teams

Manifesto

The Solo Developer's Engineering Manifesto

Engineering practice for the developer who is also the PM, QA, DevOps, and support team.

Comparison

Methodologies for Small Teams: An Honest Comparison

Scrum, Kanban, Shape Up, XP, and SSD — for teams of 1–5.

Critique

Why Standard Scrum Fails Small Teams

Scrum was designed for 5–9. Below 3, the ceremony costs more than it delivers.

Quality

Code Quality Without a QA Team: The Ratchet Principle

Encode quality in CI. The teeth of the ratchet are tests, types, and lint rules.

Releases

How Small Teams Should Think About Releases

Decouple deploy from release. Make the release a flag flip and a changelog entry.

Simplification

The Simplest Engineering Lifecycle That Actually Works

Most lifecycle complexity is scar tissue from past deployment trauma.