Built in the open
A record of what changed, why it changed, and what became more useful—not a dump of commits.
View the canonical source on GitHub →Major release · Architecture tools
Workbench became an executable architecture model
The original calculators were useful in isolation. V2 connects traffic, API capacity, cache behavior, PostgreSQL, queues, workers, retries, latency, and failure economics so a change can propagate through the whole model.
- Seven operational and failure scenarios
- First-saturation boundary detection
- Feedback-loop and coupling warnings
- Assumption confidence and production invariants
Improvement · Engineering practice
Snippets now carry production context
Code examples now document their failure policy, trade-offs, operating boundaries, and further reading. The goal is to make each snippet useful in an architecture review, not merely copyable.
Content system · Field evidence
Research and messy engineering got their own homes
Product stories capture incidents, migrations, management decisions, and the compromises that idealized architecture diagrams omit. Papershelf notes turn research into decisions, questions, and practical experiments.
Infrastructure · Discovery
The notebook became easier for humans and machines to discover
Articles gained topic tags, related reading, social sharing, and references. The site gained RSS and Atom feeds, canonical URLs, structured data, a generated sitemap, and llms.txt.
Launch
Ayushworks went public
The first version established a public engineering notebook for systems, products, research, and implementation decisions—with work spanning Saleslyt, Myotrek, healthcare systems, offline payments, and smart-agriculture infrastructure.
Release notes use an OpenChangelog-compatible GitHub source. The source stays beside the code, where changes can be reviewed and versioned.