Go back / Home
Blog
Thoughts on AI, engineering, and building products.
Blog
Thoughts on AI, engineering, and building products.
AILLMProductivity
What I Learned Testing AI Models and Harnesses for Months
A few months of trial-and-error across AI platforms, models, and harnesses. The model is not enough — harness quality, open options, and cost efficiency decide what actually ships.
10 Aug 20263 min
Read AITrendsLLM
AI Got Cheaper on the Surface — The Real Costs Moved Elsewhere
Two years ago the fear was lock-in via token pricing. On the surface AI looks cheaper than ever. Underneath, the costs moved into power, water, throttling, and ecosystem lock-in.
5 Aug 20262 min
Read AILLMEngineering
Ollama, Local LLMs, and the Hybrid Inference Playbook
Ollama made local LLMs one command away. Here is why that matters for cost, where cloud models still win, and how a hybrid local-first route actually works.
22 Jul 20262 min
Read AILLMTrends
Kimi K3 Beyond the Benchmarks
Everyone is arguing about Kimi K3 leaderboard scores. After using it alongside Claude, DeepSeek, ChatGPT, and Gemini, here is what actually mattered in real work.
18 Jul 20262 min
Read AITrendsBest Practices
AI Content: Separating Fact from Fiction
Since GPT showed up, everyone suddenly became an "AI teacher." Here is the one filter I use to tell useful AI writing from engagement bait.
14 Jul 20262 min
Read AILLMCloud
Qwen’s Latest MoE Release and What It Does to Cost-Performance
Alibaba’s latest large MoE Qwen release is interesting less as a leaderboard flex and more as a shock to the production cost-to-performance matrix — capacity, coding autonomy, and open weights.
24 Jun 20262 min
Read AILLMProductivity
Use Claude Code CLI with OpenCode Go or Zen
Claude Code only talks to Anthropic’s API. If you are on OpenCode Go or Zen, here is the open-source setup I published to bridge them with routatic-proxy.
23 Jun 20262 min
Read 
MobileAIHardware
The On-Device AI Phone Era: What’s Actually New in 2026
Smartphones are turning into personal AI computers. The real story: on-device inference, hybrid workflows, and OS-level AI that lives across your apps.
19 Jun 20262 min
Read 
SecurityAIBest Practices
Cybersecurity in 2026: AI-Driven Attacks, Identity-First Defense
AI is changing the economics of cyber attacks and defenses. Here are the trends (and habits) that matter most right now.
19 Jun 20262 min
Read 
AITrendsProductivity
Agentic AI in 2026: From Chatbots to Systems That Do Work
Agentic AI is shifting the focus from single-model demos to reliable systems: tools, memory, guardrails, and evaluation. Here’s what’s changing and what to watch.
19 Jun 20262 min
Read 
AIEngineeringLLM
From Prompt to Production: Shipping LLM Features Reliably
Most LLM demos die in staging. The gap between 'it works in the notebook' and 'it handles real users' is bigger than people expect. Here is how I close that gap.
15 Jun 20262 min
Read 
EngineeringNext.jsArchitecture
The Minimal Full-Stack Setup I Use for Every New Project
After building a lot of projects, I have settled on a setup that lets me move fast without accumulating technical debt. Here is exactly what it looks like and why.
12 Jun 20262 min
Read 
AIEngineeringProduct
Building AI Features That Actually Ship
Most AI integrations get stuck in prototype hell. Here is how to scope, build, and ship LLM-powered features into production apps without burning out your team.
10 Jun 20262 min
Read 
ReactNext.jsEngineering
Understanding React Hydration Errors (and How to Fix Them)
Hydration errors are one of the most confusing bugs in React. They look cryptic, the stack traces are unhelpful, and the cause is often non-obvious. Here is a clear mental model.
8 Jun 20262 min
Read 
Next.jsReactEngineering
Next.js Server Components: What Changed and Why It Matters
Server Components fundamentally change how you think about data fetching and component boundaries. Here is a practical guide to using them well in 2026.
5 Jun 20262 min
Read 
EngineeringAPIArchitecture
API Design Patterns That Scale: REST, tRPC, and When to Use Each
After building APIs in REST, GraphQL, and tRPC across many different products, I have a clear view of when each approach wins and where each one fails.
1 Jun 20262 min
Read 
TypeScriptEngineering
The TypeScript Patterns I Use Every Day
After years of TypeScript in production, a few patterns show up in every codebase I build. These are the ones that consistently reduce bugs and improve maintainability.
28 May 20262 min
Read AITrendsHardware
AI Is Moving Forward by Dragging Us Back to the Basics
The more advanced AI gets, the more we depend on electricity, memory, chips, and compute. Technology is a pendulum — and the bottom layers are back in the spotlight.
27 May 20263 min
Read 
SupabaseNext.jsEngineering
Supabase and Next.js: The Full-Stack Setup I Actually Use
After building several production apps with this stack, here is the exact setup, patterns, and gotchas I have settled on for Supabase with Next.js App Router.
20 May 20262 min
Read 
EngineeringBest Practices
Writing Code for the Next Developer, Not Just the Machine
Code that works is the baseline. Code that the next developer can understand, extend, and debug in six months is the actual goal. These are the habits that make the difference.
15 May 20262 min
Read Next.jsEngineeringProduct
Building a Blog Editor from Scratch with Tiptap and Next.js
I spent about a month building a custom Notion-inspired blog editor with Tiptap, Next.js, MongoDB, and Cloudinary — and why full control beat another SaaS CMS.
12 Aug 20252 min
Read