The Real Cost of AI Agents Is Task Cost, Not Token Cost
A $0.05 AI call is rarely $0.05. How to measure retries, latency, failures and human review to understand what an AI workflow really costs.
Shared schema or a database per tenant? How I choose a Django multi-tenant architecture for SaaS, with trade-offs, code and an isolation checklist.
Read More
A $0.05 AI call is rarely $0.05. How to measure retries, latency, failures and human review to understand what an AI workflow really costs.
AI makes code cheap and judgment expensive. Why the loop is now generate, review, verify, decide, and why every feature still needs an owner.
Dev servers, Docker, MCP servers and debug ports can expose files and commands. Six questions to secure the environment you build in.
How I built a calculator that plans monthly savings, timeline and affordability for home buyers, including rising prices, closing costs and couples.
Inside Macroline: live calorie and macro totals, light and double portions, goal-based presets and shareable order links, with honest data limits.
What building Savoir AI, a live platform with 80+ REST APIs, multi-tenancy and AI features, taught me about choosing Django for SaaS.
Prompt, copy, run, ship is not engineering. Habits that keep you thinking and improving while using AI coding agents every day.
Scoped permissions, sandboxes, secrets, audit logs and approvals: how to give AI agents the access they need and nothing more.
Code shows what happened; documentation explains why. A lifecycle approach to documenting decisions, requirements and assumptions.
Let's work together
I build SaaS products, scalable APIs, and thoughtful interfaces. Tell me what you have in mind and I will get back to you.