The notebook

All AI & LLMs Cloud & DevOps Industry Insights Laravel & PHP Modernisation Performance & Scale SaaS & Product Security & Compliance

All posts 113 posts

10 Apr 2026 AI & LLMs AI Agents vs Workflow Automation: A Decision Guide for SMBs A framework for when a deterministic queue/cron pipeline beats an agent and vice versa — variability, judgment, and natural-language inputs are the agent signals — with honest failure-mode a... → 10 Apr 2026 Security & Compliance Signed URLs versus the file path anyone can guess On a healthcare platform, changing one digit in a document URL opened someone else's medical report. How I replaced guessable file paths with private disks and short-lived signed URLs. → 07 Apr 2026 AI & LLMs RAG vs Long Context: When You Don't Need a Vector Database With million-token context windows and prompt caching, corpora under a few hundred pages are often cheaper and more accurate stuffed into context than retrieved — I give the cost/latency cro... → 04 Apr 2026 AI & LLMs Why RAG Fails in Production (and How to Fix Retrieval First) Roughly 73% of RAG failures are retrieval, not generation — a debugging checklist from client projects: inspect retrieved chunks before blaming the model, fix chunking (semantic over fixed-s... → 03 Apr 2026 Industry Insights Building a Fintech MVP in the Nordics: Compliance, APIs, Costs PSD2 APIs, KYC vendors, licence versus agent model — what a Nordic fintech MVP involves beyond code, and how predictably compliance inflates the budget. → 03 Apr 2026 Security & Compliance The security pass I run before every handover Before any codebase leaves my hands, it gets a fixed security pass — unbilled. Here is the actual checklist, what it keeps finding, and why it ships free with every project. → 01 Apr 2026 AI & LLMs Small Models First: Why Cheap LLMs Handle 80% of SaaS AI Features An escalation architecture I use: default every request to a Haiku/Flash-class model at cents per million tokens, escalate to a frontier model only on validation failure or low confidence —... → 29 Mar 2026 AI & LLMs Claude vs GPT vs Gemini in Production: How I Choose in 2026 A task-type decision framework instead of benchmark worship: match model class to job (extraction, chat, coding, long-doc), compute total cost per feature rather than per token, and a compar... → 26 Mar 2026 AI & LLMs Prompt Caching in 2026: When It Cuts Costs 90% and When It Won't A mechanics deep dive: how cache pricing actually works across Anthropic/OpenAI/Google, how to restructure prompts so the static prefix caches (system prompt, docs first; user input last), a... → 26 Mar 2026 Laravel & PHP Geofencing without destroying the battery Continuous GPS polling is how location apps earn one-star reviews and uninstalls. The OS-native, server-assisted geofencing setup I use instead — and the numbers behind it. → 22 Mar 2026 Security & Compliance Location verification when the prize is real money I built a live game where standing in the right place wins real money. That turns GPS — a signal designed to be trusted — into a security boundary you have to defend in layers. → 17 Mar 2026 Cloud & DevOps CI/CD for a Two-Person Team: A Pipeline That Ships in 5 Minutes Pragmatic DevOps against the over-engineering trend: the GitHub Actions pipeline I use on freelance projects — lint, test with a real database, build assets, deploy on merge — kept under fiv... →