55 articles

Insights from the work we do

Honest, practical articles from our team, what we ship, what we break, and what we learn. No fluff, just lessons from real production work.

Model Context Protocol connecting LLMs to tools
AI

Model Context Protocol (MCP): The Quiet Standard Reshaping AI Integrations

MCP is becoming the USB-C of AI tooling. Anthropic launched it, OpenAI adopted it, Google followed at I/O 2026. Here is what MCP actually is, why it matters, and how to ship with it today.

· 3 min
Claude Code CLI working alongside a developer
AI

Claude Code in Production: How We Actually Use Anthropic's CLI Day to Day

Claude Code isn't a chatbot, it's a teammate that runs in your terminal, edits your repo, and ships code. After 6 months in production, here is what works, what doesn't, and the workflow that saved us weeks.

· 3 min
AI agents architecture diagram
AI

The State of AI Agents in 2026: What Actually Works in Production

Every vendor is selling agents. Most demos fail in the real world. After a year shipping agents into actual products, here is the honest map of what works, what doesn't, and where to bet your roadmap.

· 3 min
On-device AI on smartphone chip
AI

On-Device AI in 2026: Why Edge LLMs Are Quietly Eating the Cloud's Lunch

Apple Intelligence runs locally. Gemini Nano ships on Pixel. Qualcomm Snapdragon X has a 45 TOPS NPU. The most interesting AI shift of 2026 isn't bigger models in the cloud, it is smaller, faster, more private ones on-device.

· 3 min
AI coding assistants comparison
AI

AI Coding Assistants in Production: A 2026 Comparison (Claude Code, Cursor, Copilot, Aider)

We use four major AI coding tools in production across our team. Here is where each one wins, where each one fails, and the stack we'd give a brand-new engineer in 2026.

· 3 min
AI features inside SaaS dashboard
AI

Building AI Features Into Your SaaS Product: A No-Hype Playbook for 2026

Every SaaS company is being asked to 'add AI.' Most teams cram in a chatbot and call it done. Here is the playbook we use to ship AI features that customers actually use, and pay extra for.

· 3 min
RAG vs fine-tuning architecture diagram
AI

RAG vs Fine-Tuning vs Prompting in 2026: Which One Should You Actually Use?

Most teams reach for fine-tuning when they don't need it, and avoid RAG when they should embrace it. A clear, project-tested decision tree from someone who has shipped all three.

· 3 min

Have a project? Let's
build something worth writing about.

Start a project →