ENTRY-00writing.secUI-L
Writing
Notes that come out of my own products.
Short pieces on infrastructure decisions, store releases, AI-assisted development and product scope. All of it comes from work I actually ran — not generic summaries.
Building Products with AI
From vibe coding and agent orchestration to production Lovable workflows and real AI integration: keeping product and architecture control while moving faster.
- Vibe coding is not the problem. Vibe architecture is.Fast AI-generated code is not the danger. Ownerless architecture is.
- Lessons from working with AI coding agentsAI coding agents made writing code cheaper. They did not make good decisions cheaper.
- Why GitHub should be the source of truth for AI-agent developmentWhen several agents touch one product, the hard problem is no longer generating code. It is knowing which reality is canonical.
- How Lovable, Cursor, Claude and Codex fit into one development processUsed as four interchangeable coders they create coordination debt. Given distinct roles, they can complement one another inside one engineering system.
- Not another AI chatbot: how to integrate AI into an existing productAI integration is not a chat box. It is redesigning a real job with the right context, actions and boundaries.
- Taking a Lovable project to production readinessThere is an invisible engineering layer between a working preview and a dependable production product.
From Prototype to Production
Turning a working MVP into a dependable product: production checks, modular refactoring, data migration, audits, rewrite decisions and release gates.
- 25 checks before taking an AI-built project to productionA successful build is not production readiness. I review the system from source control and auth to monitoring and rollback.
- Cleaning Preview Code: From Prototype Chaos to Modular ArchitectureA controlled way to turn AI-generated prototype code into a production codebase without throwing away what already works.
- From AI Prototype to Production Schema: Data Modeling and Migration StrategyTurning loose prototype data into a relational production model with explicit ownership, constraints, backfills, cutovers and rollback rules.
- What is an AI Product Audit, and what do I inspect?Before adding more code to a fast-grown AI-built product, I map the product and system reality.
- When should an MVP be rewritten, and when should it be rescued?Rewrite is not a cleanliness decision. It is a product-risk, migration-cost and system-changeability decision.
- Production Readiness Checklist for SaaS releasesA reusable release gate covering code, data, security, UX, operations and rollback.
SaaS Systems: Architecture, Cost and Simplicity
Field notes and decision frameworks connecting backend cost, cloud choices and product complexity as one systems problem.
- Why does a SaaS Supabase bill get out of hand?What makes a backend service expensive is rarely the pricing table — it is how your application talks to it.
- Supabase or Cloudflare? Infrastructure decisions for small productsThe useful question is not which platform is better, but which complexity the product actually needs today.
- 10 signs your startup needs simplification, not another featureGrowth does not always mean adding. Sometimes the biggest obstacle is the accumulated weight of what already exists.
- What 2.6 million Ordovia requests taught me about SaaS architectureThe request volume was not only a billing issue. It was an architectural signal that network behavior was not visible enough.
Product, systems and working notes that stand on their own.
- Why does Linear feel fast?A proof-led decomposition of Linear's creation, context preservation, keyboard, search and editing behavior without copying screens.
- From Notion and Obsidian to Agents: Building Context Architecture for AIDesigning a context system that keeps product truth, code reality, decisions and operating rules authoritative and fresh for AI agents.
- From PWA and Web to Mobile Stores: Managing Multiple Platforms from One CodebaseA practical architecture for sharing a product core across web, PWA and native stores while containing platform-specific behavior and release risk.
- Game Mechanics in SaaS Product Design: XP, Quests and Systemic HabitsDesigning gamification as a behavioral system rather than a layer of points, badges and confetti.
- How I make product decisions from idea to working SaaSA good MVP is not a product with few features. It is the smallest reliable system that tests the most important assumption.
- How I keep the system simple while building several products aloneRunning multiple products needs fewer exceptions, clearer priorities and one source of truth — not more tools.
- What it actually takes to ship a product on Google PlayProducing an Android bundle is the easy part. A real release is identity, policy, testing, store data, version discipline and production acceptance.
- Why dashboard freedom is not always good UXGiving users the right to change everything does not always make them freer. Sometimes it simply gives them another system to manage.