Model Context Protocol: Why MCP Matters for Tooling
Model Context Protocol explained by a studio that ships MCP servers: tool budgets, stdio pitfalls, OAuth rules, prompt injection risk and when to skip MCP.
11 min read
AI & Tools
Guides, tutorials and opinions on AI & Tools from the team behind Canvas.
Prompt Engineering for Developers: Patterns That Survive Model Updates
Prompt engineering patterns that survive model updates: strict JSON schemas, golden set evals, snapshot pinning and the few-shot habits that break on upgrade.
13 min read
Choosing an LLM API: Cost, Latency and Lock-in
A practical llm api comparison for 2026: real cost drivers, latency maths, OpenAI vs Anthropic API differences, prompt caching and how to avoid lock-in.
12 min read
AI and SEO: What Search Engines Reward and What They Penalise
What Google actually rewards and penalises in AI content SEO: scaled content abuse, information gain, AI Overview citations, crawler control and real
11 min read
Fine-Tuning: The Cases Where It Still Wins
When fine tuning an LLM still beats prompting and RAG in 2026: the four winning cases, LoRA configs that work, dataset rules and the real break-even maths.
13 min read
AI Coding Assistants Compared: Where Each One Actually Helps
A practitioner’s 2026 comparison of every major ai coding assistant: Copilot, Cursor, Claude Code, Aider and more. Where each one helps, and where it fails.
12 min read
Using AI to Write Code You Can Actually Maintain
How to keep ai generated code quality high in 2026: context files, 300-line diff caps, machine-enforced rules and the code review AI can’t do for you.
10 min read
Building an Internal Knowledge Assistant on Your Own Docs
How to build an internal knowledge assistant on your own docs: eval sets first, heading-aware chunking, hybrid pgvector plus full-text retrieval, reranking
13 min read
