Writing
Field notes on building
conversational AI.
Essays on prompts, evals, and orchestration, for teams building AI.
Not sure where to start?Take the 2-minute quiz. I'll point you to the essay for you.What kind of AI collaborator are you?Why People Don't Read Anymore
The share of Americans reading for pleasure daily fell from 28% to 16% in 20 years, while book sales rose. What the data says, which viral explanations fall apart, and what brings readers back. With charts.
Breaking into AI in 2026
Ten AI career paths. For each one: the education that helps, the personality that fits, and the tech to learn first. Backed by research, written in plain language.
The LLM Evaluation Framework
Building the eval harness is half the job. The other half is deciding what it gates on: containment, latency, satisfaction, and cost, held as one vector instead of a single number that flatters whoever is releasing. The operating manual companion to Evaluations are the product.
What Toy Story 5 Gets Right About Kids, Tech, and AI
I took my family to see Pixar's new movie, and a toy and a tablet explained my whole career back to me. An opinion piece on why the film's message about technology is the one we keep getting wrong about AI. Spoilers inside.
Let the AI Interview You, Then Make a Plan
The biggest gains I see come from two habits you can package as skills. Let the AI interview you until the goal is clear. Then have it plan the work before it runs.
Ecommerce AI Hit an Inflection Point
I pulled every ecommerce-AI project on GitHub from 2024 to early 2026 and ran machine learning over it. A small category became a crowd. The essay covers where it was, where it is, and where it is going.
How Google's AI Overviews Work, and How to Earn a Citation
Why Google still matters in the ChatGPT era, what happens behind the AI Overview box, and how to write for it.
Evaluations are the product
Reliable AI teams build the evaluation harness first. They commit it to version control and wire it into every release.
Context engineering: the prompt is one piece
Strong results from 2026-era models come from context architecture. The design artifact is what the system assembles around every prompt, turn after turn.
The model works. The handoff fails.
When AI conversations go wrong, the model is the easy thing to blame. The failure I keep seeing is in the handoff between people and the model.
Content Engineering in 2026
Content engineering keeps AI answers reliable. Here is what the work looks like.
Weekly on ICX
New essays every week on the ICX blog.
I publish longer pieces on conversational AI, orchestration, evaluation, and the work behind client engagements over at Intelligent CX Consulting. Fresh writing drops every week.
Work with me
Two ways to work with me.
Consulting through ICXA few consulting engagements each quarter through Intelligent CX Consulting . Start at services.
Hiring ChristiOpen to conversational AI, AI product, and applied AI roles. View my résumé.