Blog
Notes on AI, software architecture and engineering leadership, from building and shipping real products.
Showing 14 of 14 articles
When an AI Company Goes Bust, Who Gets Your Chats? Lessons from Character.ai's Bankruptcy
Character.ai's bankruptcy shows that private chatbot conversations can become a sellable asset. What users and AI builders should do about it.
China's AI Stock Bubble Has Burst: What It Means for People Building with AI
China's AI-heavy STAR 50 index fell about 40% and wiped out $480 billion. What happened, and five lessons for teams building AI products.
Coding Agents in Your CI/CD Pipeline: How to Let AI Open Pull Requests Safely
How to run AI coding agents inside CI/CD with least-privilege tokens, sandboxed runners, no production secrets and required human review.
Building AI Support Bots That Actually Help: Retrieval, Guardrails and Human Handoff
A practical architecture for customer support bots: answers from approved content, safe tools, guardrails, fast human handoff and the right metrics.
Deno Is Joining Cloudflare: What It Means for Supabase Edge Functions
Deno Deploy is shutting down and the Deno runtime enters maintenance. What that means for Supabase Edge Functions and how to keep your code portable.
The GEO Playbook: How to Get Your Content Cited by ChatGPT, Gemini and Perplexity
A practical, step-by-step guide to making your website easy for AI assistants to find, trust and cite — plus a 30-day starter plan.
SEO vs GEO: How LLM Search Is Changing the Way People Find You
AI assistants now answer questions directly. What GEO (Generative Engine Optimization) is, how it differs from SEO, and why your site needs both.
Small Language Models and Model Routing: Cut AI Costs Without Losing Quality
Why production teams are moving to the smallest model that does the job, how model routing works, and how to switch safely using evals.
LLM Evals: How to Test AI Features Before Your Users Do
A practical guide to building eval sets, choosing graders and making evals a release gate for prompts, models and RAG pipelines.
AI Coding Agents and the Engineering Team: What Changes for Leads
Coding agents can now draft whole pull requests. What genuinely changes for engineering teams, what doesn't, and a practical playbook for leads.
Model Context Protocol (MCP) Explained: The Standard Way to Connect AI to Your Tools
MCP is becoming the USB-C of AI applications. Here's how it works, what servers expose, why it matters for businesses, and how to use it securely.
Agentic AI in Production: From Chatbots to Agents That Get Work Done
What makes an AI system an agent, where agents work well today, and five lessons for putting one into production safely.
Semantic Candidate Matching with Apache Solr and Vector Embeddings
How we moved recruitment search beyond keyword matching by combining LLM-based CV parsing, vector embeddings and Apache Solr — and what to watch out for.
From Senior Developer to Engineering Team Lead: What Actually Changes
Lessons from leading engineering teams while staying hands-on: code reviews, unblocking people, saying no, and measuring success by the team's output.