Retire the ritual

An old process can acquire a remarkable amount of protection: a template, an owner, a quarterly reminder, and now an AI agent. This issue asks what earns that protection. An effort to define AI value meets arguments for redesigning work and protecting the judgment that guides it. We think a useful place to start is the report that keeps arriving long after anyone remembers what decision it supports.

Defining value means choosing what counts

The Linux Foundation launched the Tokenomics Foundation to develop shared standards for AI economics, including how spending relates to business outcomes.

Our take: We like the ambition to agree on what value means. The proposed roadmap includes measuring work completed without human involvement, which deserves a harder question: did anyone need that work completed? A system that produces an unwanted report all by itself has achieved an impressive degree of independence from its audience.

Don’t automate a process you ought to retire

McKinsey’s argument for escaping the HR pilot trap warns that isolated AI projects can perfect steps a redesigned workflow would eliminate.

Our take: A faster approval maze is still an approval maze. We would borrow the article’s question about what to stop doing before buying its whole picture of HR in 2030. Teams can make that decision now: does this report, handoff, or review actually help someone act? Automating the ritual can make it harder to question later.

Keep the first diagnosis yours

David Lancefield’s HBR essay on leadership and AI draws on his advisory work with executives to argue for retaining judgment, voice, and direct contact while delegating bounded tasks.

Our take: A review step at the end cannot recover a question nobody asked at the beginning. We would bring a short diagnosis into the room before reading the generated briefing, then use AI to challenge it. That changes the tool’s job from setting the agenda to testing it. And keep actual customer and employee conversations: a tidy digest is a poor substitute for hearing someone hesitate.

One useful move

Have a recurring meeting agenda that’s mostly updates? Give it to your usual AI tool with the notes you already have:

Meeting updates: {{meeting_updates}} Turn these updates into a short note I can share before the meeting. Use only the information provided. At the end, list the questions or decisions that still need a conversation. Flag anything the notes don’t answer.

Check the note, fill in any essential gap, and use it to give the conversation a clearer purpose.

Something fun

Ask your chatbot for a deeply bureaucratic fairy tale:

A knight must obtain five signatures to borrow a pencil. Write a short fairy tale about the quest. At the final desk, a clerk discovers the form was abolished 80 years ago. Give the knight an appropriately underwhelming reward.

We hope it’s an eraser. The pencil requires a separate quest.

Built in the past two weeks

We documented a repeatable production process for turning workshop and coaching recordings into edited videos, recaps, and follow-up resources. It distinguishes a member recording from a short public preview and carries the work from transcript and cut plan through checking the exported file. The new SOP gives each kind of recording its own editing and recap requirements.