Long-form essays from AK-DataGeek on cloud warehouse migrations, bitemporal modelling, Hermes Agent and Claude Code agentic workflows, and the practical tradeoffs we make in production. No fluff, no hype — just what we've learned shipping systems for regulated industries.
Self-improving agents sound like hype until they start remembering your pipeline failures, extracting reusable skills, and cutting ETL recovery from hours to minutes. A practical field report on Hermes Agent, persistent memory, MCP tools, and what actually works in production.
→Why Oracle migrations are accelerating in 2026, the cost-benefit reality nobody puts in slide decks, and a five-phase playbook drawn from real engagements. Plus where AI augmentation actually saves time — and where it makes things worse.
→Backtests that look profitable in the lab and lose money in production are usually betraying one bug: point-in-time data leak. Bitemporal modelling is how we fix it. A practitioner's guide with implementations in BigQuery and Snowflake.
→A boutique consultancy delivering enterprise-scale work with one engineer sounds like a magic trick. It isn't. Here's exactly how Claude Code, MCP servers and CLAUDE.md files reshape the economics of senior consulting — and where we still won't use AI.
→