From Client Work to Products: MetaGrid, BulkFix, VisaChamp
How MetaGrid, BulkFix, and VisaChamp grew from client work: spot repeated pain, ship a thin tool, and do not wait for a platform—lessons from Karachi.
What a Context Window Actually Means for Your Product
What an LLM context window means for your product: long PDFs, chat history tax, silent quality drop at the edges, and why RAG beats stuffing the handbook.
A Speed Budget for Shopify (Core Web Vitals Without the Theatre)
Set a Shopify speed budget for Core Web Vitals without the theatre: weight limits per template, what to cut before a new theme, and practical LCP/INP habits.
Multi-Model Routing: Cheap First, Expensive Only When Needed
Multi-model routing for LLMs: classify first, send easy work to cheap models, promote failures—and keep a cost ceiling on Shopify bots and catalog pipelines.
Headless Shopify: When Hydrogen Is Worth the Migration
When Shopify Hydrogen is worth a headless migration: theme ceiling vs cost, go/no-go checklist, and what merchants underestimate.
When a Cheaper Model Is the Right Model
When a cheaper LLM is the right model: draft-then-promote with mini/Flash first, flagship only on failures, and how to measure the blend.
Saying No to “Just Add AI” Without Losing the Work
How I say no to “just add AI” without losing the work: reframe into a scoped job, five cost numbers, and a monthly ceiling in the SOW.
Structured Outputs: Stop Parsing Prose, Start Returning JSON
Structured outputs for LLMs: use JSON schemas and validators so free-text stops breaking your Shopify and automation pipelines.
Estimate in Systems, Not Hours: A 6-Year Lesson
A 6-year lesson: estimate in systems, not hours. Price retries, feeds, admin UI, and AI meters—not the happy-path demo—when scoping Shopify and AI work.
Logging LLM Calls Without Leaking Customer Data
How to log LLM calls without leaking customer data: PII redaction, sampled traces, debug-enough retention, and Shopify support-bot gotchas.