The ColdFusion guides, and what else I publish.

Two different things live on this page, in the order that matters: the ColdFusion and CFML guides written for this site, then essays published elsewhere on applied AI and AWS engineering.

Guides written for this site.

These are the questions I answer most often on assessment calls. The upgrade guide, migration decision framework, and security-audit scope are all published. If your situation is more specific than a guide can be, the assessment call is free.

Adobe ColdFusion to Lucee: a decision framework · published

When to stay on Adobe, when Lucee makes sense, and where BoxLang fits — compared on support model, licensing, engine-specific dependencies, deployment fit, and codebase compatibility. Read it →

The ColdFusion security audit: what it actually covers · published as a service page

A transparent walkthrough of what gets checked, why, and what the written deliverable contains — so you know exactly what you're buying before you buy it. See what the audit covers →

Upgrading ColdFusion 2018 and 2021 safely · published

A practical inventory, parallel-build method, compatibility checklist, testing gates, and rehearsed rollback — including the controls to use if you cannot upgrade this quarter. Read the ColdFusion upgrade guide →

Have a question the guides do not answer? Send the environment details or book the free technical assessment.

Applied AI lab notes.

Ten published pieces, several of them in AWS in Plain English — a Medium publication, which means an editor other than me decided they were worth running. They are worth being clear about: these are not ColdFusion or Lucee articles. They're field notes on building and operating AI systems on AWS, and on where AI-assisted engineering goes wrong.

They're here because the most common and most reasonable question about a solo consultant is whether he's still building things. These are the receipts — and they sit below the ColdFusion material rather than above it, which is the right way round for anyone arriving here about CFML.

Start here

Scaling AI Memory: My Messy Journey Building an 8-Layer System for Claude Code

How the memory architecture behind AI-assisted development actually gets built — and why "messy" is the accurate word for the first eight attempts.

When AI Lied to Me for Two Hours (And I Totally Fell for It)

A working account of confidently wrong AI output surviving two hours of review. This is the essay behind every verification check I now build into legacy-code work.

Migrating to AWS Bedrock AgentCore: The Journey Continues

A real migration onto a managed agent platform, written while doing it — including the parts that didn't go to plan.

From Accidental Grief Counselor to Digital Jung: Deploying a Jungian Philosophical Companion on OpenCLAW

Deploying and securing a self-hosted AI agent, with a full section on trust boundaries, tool policy, sandboxing, and treating inbound content as data rather than instructions.

Everything else

The Mirror and the Crucible

On testing what an AI system actually does under pressure, rather than what its description says it does.

The $200 AI Tip Trick Is Dumb… and It Works Anyway

A prompt-engineering result that has no business working, tested honestly instead of dismissed.

The Night My Windows 11 Laptop Threw in the Towel (And How an EC2 Instance Saved My AI Coding Flow)

Practical AWS problem-solving at 2am, which is when most infrastructure decisions are actually made.

When AI Gets Stuck

Recognising the failure mode where a model stops making progress and starts producing plausible motion — and what to do about it.

Enhancing AI Collaboration through Structured Post-Mortems

Applying ordinary engineering post-mortem discipline to AI-assisted work, so the same mistake stops recurring.

I Gave My Friend ChatGPT for Job Hunting. It Became Her Grief Counselor Instead.

What people actually do with these tools once nobody is supervising the use case.

Read more on Medium

Book a free assessment and get answers now.

I'll review your ColdFusion version, licensing, and vulnerability exposure in a 30-minute call.

Book a free assessment