Notes
Since 2023 I have been writing down what an AI transformation asks of leadership. Each note shows when it first appeared and when I last revised it.
2026
- AI answers by language: the market in the query decidesIn global categories, naming the market pulls in local sources for AI search, the language doesn't. What 15 Perplexity queries showed, plus a 20-minute check.
- Shutting down an AI system: keeping it is a decision tooShut down an AI system when usage, quality, upkeep or a better alternative say so. Four checks for every live system, and a shutdown order that strands nothing.
- Running AI: catch failures, respond under controlYou only catch AI failures if you check availability, output format and business quality separately. Responding under control means putting a limit on retries.
- AI cost management: know the cost per call or fly blindYour AI provider's invoice won't tell you which workflow is spending the money. How to track AI costs per call, budget each app and weigh cost against value.
- AI quality assurance: a glance is not a reviewCheck AI output in three separate passes: format, source, expert judgment. The sender stays responsible. With a review table and a rule for model switches.
- llms.txt: an overview with clear limitsAn llms.txt links a site's key pages with short descriptions for AI systems. It helps them find their way around, but it does not guarantee any citations.
- ChatGPT or Claude: judge the model by the taskWhether ChatGPT or Claude fits better shows only in a test on your own task, with scoring set in advance and the full cost of running it in production counted.
- AI prompts as part of the documentationPrompts need documenting because they hold business rules. Record the goal, each version with its reason, test examples and who owns the rule in the business.
- Running AI in production: the pilot is the easy partAn AI pilot proves little. Production starts where others build on the output. How we take AI systems into production, with six questions to ask before you go.
- Build vs. buy AI: only the build got cheaperBuilding AI got cheap, running it didn't. When building your own AI pays off and when buying is smarter, with a decision table for the leadership team.
- OKRs in practice: goals that steer the workdayOKRs only steer the workday when key results measure impact instead of counting activity. How I lead with few goals as COO, and where the method is a poor fit.
2024
- Measuring AI ROI: time saved isn't money savedMeasure AI ROI in hours per deliverable from your own time tracking, review included. Time saved only turns into money once the team puts it to work.
- AI infrastructure: a pile of tools is not a systemAI infrastructure starts with users, permissions and access for everyone, then tools. Your own layer keeps models swappable and books the cost of every call.
- AI change management: your team is rarely the problemAI transformation fails more often for lack of learning time than because of the team. Why pushback is a review brief, and five questions before any new tool.
- AI pilot to production: pick the tool lastAn AI pilot is ready for production once it runs without the pilot team. Plan back from the final handoff: sources, review, access. Pick the tool last.
2023
- GPT-4 and SEO: looking back at how the work changedIn SEO work, GPT-4 takes over routine work: preparing data, grouping queries, drafting. Checking and deciding stays with the editors. A view from October 2023.
- AI adoption: stop treating it as an experimentAI adoption is a leadership decision. A subscription for everyone won't do it: settle data, access and review first, then run a focused trial on one workflow.