May 23, 2026
Edge vs Node for LLM and DB routes: limits and split patterns
What breaks in edge runtimes with database drivers and LLM calls, and how to split routes once you are already running in production.
Notes on product, AI automation, and technical decisions.
No editorial gloss.
May 20, 2026
Stripe webhooks: idempotency, duplicates, and concurrency
A minimal schema and transactional handler for duplicate, out-of-order, or at-least-once Stripe webhooks, without assuming each event is unique.
May 17, 2026
Lightweight RAG: embeddings, SQLite, and when to skip a vector DB
RAG tradeoffs at small scale: vector size, storage, and when SQLite is enough before you pay for a dedicated vector database in production.
May 14, 2026
Minimal tracing and costs for AI pipelines in production
Which metrics actually matter for LLM pipeline cost and reliability in production, before you buy an enterprise observability stack.
May 11, 2026
Structured LLM output with Zod: streaming, validation, and recovery
TypeScript patterns to validate structured LLM output, handle streaming, and recover when JSON or the schema fails once you are in production.
Mar 8, 2026
Cron without a server: GitHub Actions calling your API (scheduled webhook)
Serverless cron: a scheduled GitHub Actions workflow, an authenticated POST to your app, timing and minute limits. A pattern, not a tool gallery.
Mar 7, 2026
llms.txt: what it is good for (and what is not proven)
llms.txt is a machine-readable summary, not a proven ranking lever. What it is for, the limits, and how we serve it from Nitro rather than public/.
Feb 28, 2026
The Hidden Cost of SaaS Tools: When to Develop Your Own Solution
Stacking cheap monthly tools becomes an opaque fixed cost. When a SaaS pile stops being cheaper than building the piece you actually use.
Feb 21, 2026
From Excel to Automated Dashboard: The Complete Guide
Excel is fine until updating sheets eats more time than decisions. How to know when to migrate without losing data or freezing operations.
Feb 14, 2026
What a digital product costs (without a made-up number)
A cost framework: scope, integrations, risk, ownership. Why a fixed low quote is often a red flag, and what to bring to a call.