Blog
Field notes on OSINT, security checks, SEO, and safer hiring and vendor due diligence.
-
Why interface design is a security control in blockchain and AI
2026-08-23
A quote highlighted by Simon Willison points to a practical lesson for blockchain and AI teams: developer experience is a security control. Good interfaces, visible state, narrow authority, and failure-first testing reduce costly mistakes.
-
What LLM CLI Releases Teach You About Production AI
2026-08-22
A small LLM CLI release points to a bigger engineering shift. Once model calls move into scripts and services, prompt execution becomes infrastructure, with versioning, contracts, and security boundaries to test.
-
What site: at scale means for AI retrieval systems
2026-08-21
AI search systems are leaning harder on query operators like `site:`. For teams building RAG, assistants, or documentation platforms, this changes how retrieval behaves and how site architecture affects answer quality.
-
Go 1.27: How to assess a release before you upgrade
2026-08-20
Go 1.27 is a prompt to review runtime behavior, standard library boundaries, and build reproducibility in your own services. A safe upgrade depends on benchmarks, contract tests, and staged rollout, not on release-post headlines.
-
How to evaluate an open source AI language in production
2026-08-19
Mojo becoming open source matters less as a headline and more as a chance to inspect the compiler, runtime, and interop story yourself. Here is how to evaluate a new AI-focused language before it enters your production stack.
-
Markdown SVG support changes your render security model
2026-08-17
SVG support in Markdown improves developer experience, but it also expands the browser attack surface. If your app renders user or model-generated Markdown, review SVG as active content and test the full render pipeline, not one sanitizer in isolation.
-
CORS Proxies in Browser-Based AI Systems
2026-08-16
Browser-facing AI apps often add a CORS proxy to make front-end calls work. The real issue is not CORS itself, but where trust sits, which credentials reach the browser, and how tightly your proxy enforces origin, route, and upstream policy.
-
What a Bird Registry Teaches You About Blockchain Data Design
2026-08-15
A short field note about a tagged seabird points to a deeper engineering pattern: distributed records need stable identifiers, provenance, late-data handling, and public verification. Those same design choices shape resilient blockchain systems.
-
Using SQLite generated columns in AI and Go systems
2026-08-14
SQLite generated columns are a practical way to normalize JSON-heavy data in AI and Go systems. They help you keep raw records intact, define one source of truth for derived fields, and build indexes around real query paths.
-
What a New Hosted Model Release Should Change in Your AI Stack
2026-08-13
A new hosted model release is a good prompt to review the parts of your AI stack most likely to fail in production. Focus on gateway routing, prompt isolation, drift detection, and plain Go interfaces around changing model backends.