AI workflow 2026 – 2 years later, what works, what doesn’t, Claude 200K in practice
Two years after I first wrote about AI coding tools, the landscape has changed completely. Context windows went from 8K to 200K+ tokens. Models that were unreliable on Magento code are now genuinely useful. Cursor IDE turned AI from a suggestion machine into a pair programmer. I revisit my 2024 assessment with two years of production experience and concrete workflow changes.
- Published in PHP
OpenSearch 3.x vector search – embeddings via Ollama, k-NN, hybrid search
OpenSearch 3.x adds neural/vector search capabilities that were previously only available in specialised vector databases. Combined with Ollama running locally for embedding generation, you can add semantic product search to Magento 2 without external API dependencies. I show the complete implementation: generating embeddings, storing k-NN vectors, hybrid search combining text and semantic scoring.
- Published in Magento / Adobe Commerce
Interpreter pattern – own discount rule grammar, parser, expression tree
Interpreter is the GoF pattern for implementing a mini-language. Instead of hardcoding discount rules, you define a grammar and an interpreter that evaluates expressions at runtime. This makes complex pricing rules configurable without code deployments. I build a complete discount rule language in PHP: grammar definition, recursive descent parser, AST, and evaluation.
- Published in PHP, Wzorce projektowe
AI-assisted SQL optimisation – LLM + EXPLAIN + Blackfire, 5x speedup
SQL optimisation has always required human expertise, but the workflow has changed. I now use an LLM as a pair programmer when reading EXPLAIN output – it translates the cryptic rows into plain English and suggests indexes. Blackfire shows me where the time goes. Together they cut a query from 4.2 seconds to 0.8 seconds in an afternoon. I show the complete workflow with real examples.
- Published in Magento / Adobe Commerce, PHP