The AI Playbook presents
RE+SHAPE
How should this get done now that AI exists?
Most AI advice starts with the technology. RE+SHAPE starts with the work. Find what is worth changing, question why it works the way it does, and decide the right role for people, intelligence and automation.
Keith Eddleman
CTO at Q1Media · Author of The AI Playbook
Don't just add AI to the way you work. RE+SHAPE how the work gets done.
I've been working on a question: how should work get done now that AI exists? It led me to a framework I'm calling RE+SHAPE.
It is not an automation maturity model. Run a piece of work through it and the right answer might be to eliminate it, to leave it exactly as it is, to have a person and a machine do it together, to hand it over entirely, or to let it run on its own. All five are real answers.
The goal isn't more AI. The goal is better work.
The framework
RE finds the work worth changing. SHAPE determines how it should change.
RE
Find the work worth changing
Reveal
Surface the candidates. Friction, repetition, wasted attention, recurring meetings, manual handoffs, copy-and-paste, spreadsheets standing in for processes, and valuable work that used to be too expensive to do at all.
Reveal produces hypotheses, not verdicts.
Examine
Decide whether the work is actually worth changing. Why does it exist, who benefits, where do judgment and relationships matter? A valid conclusion is: do not change this.
A valid conclusion is “do not change this.”
SHAPE
Determine how it should change
Specify
Define the outcome, not the mechanism. Say what finished looks like without describing the way it happens to be done today.
Can the outcome be stated without prescribing the current mechanism or a premature AI solution?
Hunt
Question the workflow before adding anything. Why does each step exist? Why is a human touching it? Classify every step: delete, reduce, preserve, elevate.
Have we simplified before adding AI?
Augment
Put intelligence where a person plus a machine genuinely beats either alone. Information should reach a person already sorted, so their attention is an escalation path rather than a processing layer.
Is the augmented workflow good enough that the human will actually change the old behavior?
Put AI to Work
Transfer responsibility. Give AI a job, not a prompt: a mission, real context, examples of the judgment you want, the sources that count, and a definition of done.
Does the human trust the delegated work without routinely redoing or checking the original?
Execute
Transfer initiative. The work starts because the situation calls for it, not because somebody remembered to ask. Triggers, authority, boundaries, approvals and escalation all get written down.
P gives AI responsibility. E gives AI initiative.
You don't win by reaching E.
Sometimes the right answer is to eliminate the work. Sometimes it is to leave it alone. Sometimes a person and a machine together is best. Sometimes AI should own the job outright.
Autonomy is earned, not assumed.
Case study
I Didn't Automate My Standup. I Rethought Why I Had to Attend It.
Four developer standups a week, 45 to 60 minutes each. Roughly 80% of it is implementation-level status that does not need a CTO in the room. The obvious AI answer would be to record and summarize the meeting.
Before
- 4 standups a week
- Consume all status
- Catch the 20% that matters
After
- Updates that already exist
- An AI attention layer
- Decisions, blockers, initiative movement
- A higher-value human meeting
Reveal
A recurring workflow consumes roughly 3–4 hours a week of CTO attention. That makes it a high-leverage candidate for examination.
Examine
The meeting is not only information transfer. It also provides coaching, leadership presence, connection, accountability, rapid decisions, demos, and coordination with an India-based team. Canceling it blindly would destroy value.
Specify
Maintain CTO-level situational awareness while minimizing implementation-level information consumption, without delaying decisions that could cost the offshore team a work cycle.
Hunt
Delete the routine verbal status recitation. Reduce CTO attendance and repeated transmission. Preserve connection, accountability, leadership access and coaching. Elevate blockers, decisions, demos, architecture and priority changes.
Augment
A daily development update already exists and already contains enough raw information, so nobody writes a new report. AI evaluates it against Keith's priorities and produces only what requires attention — and may say nothing requires your attention today.
Put AI to Work
A persistent role: CTO Development Intelligence Analyst. Its mission is to maintain CTO-level awareness while shielding Keith from implementation noise. It passes when Keith no longer feels compelled to read the original report just to make sure.
Execute
When the update arrives the analyst preserves the source, extracts events, compares them against initiative state, updates the chronology, detects meaningful change or prolonged silence, and decides whether Keith should be interrupted at all.
The fifteen-second brief
Nothing requires your decision today. Margin-alert work is progressing through QA; QA found a navigation defect in the below-40% margin email. Purdue's earlier-data request and Google Keywords pipeline work are worth watching, but no ingestion failure is reported. Console AI's “What's New” feature is in UAT. No commission, BDR, Events, or other major AI-initiative activity was reported today.
AI consumes status. Keith shows up for the humans.
The goal wasn't to automate the standup. The goal was to rethink why the standup existed.
Core principles
What the framework keeps telling you.
- Stop making humans find what matters.
- Preserve the value. Eliminate the mechanism.
- Give AI a job, not a prompt.
- Never automate work you haven't learned to delegate.
- Connected doesn't mean authoritative.
- Automate last.
The 30 Plays
Thirty behavior changes, not thirty AI features.
Simplification
Preserve the Value. Eliminate the Mechanism.
You radically change or remove a mechanism without losing the value it was meant to provide.
Intelligence
Never Read Everything Again
Your instinct changes from “I need to read all this” to “I need to interrogate this body of information.”
Intelligence
Stop Making Humans Find the Signal
You no longer consume an information stream to discover what matters. You enter the process when something matters.
Responsibility
Give AI a Job, Not a Prompt
You stop re-explaining a recurring task from scratch.
Responsibility
Define the Source of Truth Before Connecting Everything
AI knows not only where it can retrieve an answer but which source has authority.
Autonomy
Give AI Initiative
Useful work begins because the situation requires it.
You made it this far. You're clearly my kind of reader.
Join enterprise technology leaders and practitioners.
Free. No spam. Unsubscribe any time.