Notes on building.
Guides on mobile apps, web platforms and AI products. Cost breakdowns, stack comparisons and how we ship.
Feature Flags for AI: Safely Rolling Out LLM Features in Production
Shipping AI features without feature flags is like deploying to production without a rollback plan. You will eventually push a bad prompt, a hallucinating model, or a cost-exploding configuration to 100% of your users at once. Here is how to stop doing that.
Technology · Apr 24, 2026 · 14 min read
Apple Intelligence vs Android AI: Building for Both Platforms 2026
Apple and Google have taken wildly different approaches to on-device AI, and building for both without duplicating your entire codebase is harder than either company wants you to believe. Here is what actually works.
Technology · Apr 24, 2026 · 14 min read
Expo Router vs Next.js App Router: Universal App Architecture 2026
Both Expo Router and Next.js App Router use file-based routing with React Server Components, but they solve fundamentally different problems. Choosing between them for a universal app determines your deployment targets, performance ceilings, and team velocity for years.
Technology · Apr 24, 2026 · 14 min read
The CEO's Guide to AI Coding Agents: ROI, Risks, and Team Impact
The 2-3x productivity claims around AI coding agents are directionally correct but dangerously incomplete. This guide breaks down what actually changes in your engineering org, where the real risks hide, and how to build an adoption plan that does not blow up your team culture.
AI & strategy · Apr 24, 2026 · 14 min read
AI for Vertical SaaS: Building Domain-Specific Intelligence 2026
Horizontal AI tools compete on price. Vertical SaaS with embedded domain intelligence competes on value, and that difference shows up in 3-5x higher revenue multiples. Here is the playbook for building it.
AI & strategy · Apr 24, 2026 · 14 min read
Building an AI Company vs a Company That Uses AI: Strategy Guide
The distinction between building an AI company and building a company that uses AI is not semantic. It determines your valuation multiple, your hiring plan, your technical architecture, and ultimately whether you win or lose your market.
AI & strategy · Apr 24, 2026 · 14 min read
AI Agent Security: Preventing Prompt Injection and Data Leaks
AI agents that can browse the web, call APIs, and execute code are powerful, but every new capability is a new attack surface. If you are shipping agents without layered prompt injection defenses and strict data boundaries, you are one crafted input away from a breach.
AI & strategy · Apr 24, 2026 · 14 min read
Trigger.dev v3 vs BullMQ vs Graphile Worker: Background Jobs
Trigger.dev v3 gives you managed infrastructure and zero-ops. BullMQ gives you battle-tested Redis queues. Graphile Worker keeps everything in Postgres. Choosing wrong costs you months.
Technology · Apr 23, 2026 · 14 min read
When to Replace Your SaaS Tool Stack With Custom AI Agents
The average mid-stage startup spends $50K or more per year on SaaS subscriptions, and half those tools do work an AI agent could handle for a fraction of the cost. Here is how to decide what to replace, what to keep, and how to avoid costly mistakes.
AI & strategy · Apr 23, 2026 · 14 min read
How Much Does It Cost to Migrate From Heroku to AWS in 2026?
Heroku's pricing reset in 2022 forced thousands of teams to rethink their hosting. If you are evaluating a move to AWS, here is what it actually costs, what breaks during migration, and whether AWS is even the right destination.
Cost & planning · Apr 23, 2026 · 14 min read
How to Build an AI-First Internal Knowledge Base for Your Team
Your team wastes hours every week searching for answers buried in Slack threads, Google Docs, and Confluence pages. An AI-powered internal knowledge base fixes this by actually understanding questions and pulling the right answers from every source your team uses.
How to build · Apr 23, 2026 · 14 min read
How to Build a Weather Data and Hyperlocal Forecast App in 2026
Most weather apps are just thin wrappers around the same stale government data. If you want to build something people actually pay for, you need hyperlocal ML forecasts, sub-minute severe weather alerts, and a data pipeline that does not fall over during a hurricane.
How to build · Apr 23, 2026 · 14 min read