Field notes
Practical AI engineering writing for founders and technical teams who need sharper AI strategy, better product delivery, and a developer who can turn business automation ideas into shipped systems.
Claude Code and I diagnosed why calls were dropping on production voice bots. The fix it built passed every test and missed the cause we had found, so 33 calls dropped the next night. How that happened, and why the person who answers for it is the reason software engineering isn’t dead.
Google sign-in for SuiteCRM 8: when the built-in SAML sign-in is enough, and what a custom build adds when the same login must also grant access to Gmail and Google Calendar.
A production war story: the root disk hit 94% with hours to spare before an army of voice agents went live. How I moved a 136 GB InnoDB table to a spare volume with zero data loss, using MySQL 8.0 innodb_directories, not a symlink.
The real skill stack behind AI automation: no-code tools, programming, SQL, system design, security, prompt engineering, LLM observability, and RAG, plus the order to learn them in.
A SuiteCRM case study: converting accepted quotes to orders, invoicing order deltas, and detecting supplemental invoices and credit notes.
A SuiteCRM migration case study: Docker, Google SSO, calendar sync, OAuth, multilingual fields, and production-ready CRM infrastructure, plus a SuiteCRM 7 to 8 checklist.
How AI engineering, foundation models, and automation create measurable ROI, plus where to invest first and how to ship practical AI.