High Agency, Low Tolerance Model
Pair proactive experimentation with intolerance for fixable friction.
- Difficulty
- Advanced
- Time to result
- ~ongoing to results
- Steps
- 6
- Confidence
- 95%
The High Agency, Low Tolerance Model describes effective AI adopters through two reinforcing traits. High agency means being curious, trying things, and building capabilities that were previously inaccessible rather than waiting for complete instructions. Low tolerance means refusing to accept avoidable meetings, clumsy workflows, or poor outputs when they can now be improved quickly. Together, the traits create a loop: curiosity reveals what is newly possible, dissatisfaction identifies worthwhile targets, and rapid AI-assisted building produces improvements in minutes or hours. The model can serve both as a personal development lens and a team scorecard. Its purpose is not reckless action; experimentation should remain bounded by authority, risk, and accountability.
Origin
Extracted from Marketing Against The Grain.
Core principles
- 01Act before certainty arrives
- 02Use curiosity to discover newly possible work
- 03Treat fixable friction as a prompt to improve the system
- 04Exploit AI's shorter build cycles
- 05Judge agency and tolerance together
How to run it
- 1
Assess agency
Look for evidence that a person explores, experiments, and creates without requiring exhaustive direction.
Pro tip Evaluate completed experiments rather than self-described enthusiasm.
Watch out High activity without useful outcomes is not necessarily high agency.
- 2
Assess tolerance
Observe whether the person accepts recurring friction or actively tries to improve it.
Pro tip Ask which broken workflow they most recently changed.
Watch out Low tolerance should target systems and outputs, not become hostility toward colleagues.
- 3
Find the intersection
Prioritize opportunities where proactive ability and dissatisfaction with the status quo reinforce each other.
Pro tip Start with friction that can be improved through a short, reversible experiment.
Watch out Do not encourage unilateral changes to high-risk or governed processes.
- 4
Build quickly
Use AI to prototype a better process, tool, analysis, or artifact within minutes or hours where feasible.
Pro tip Make the smallest version that can prove the improvement.
Watch out Speed does not remove the need to test accuracy and safety.
- 5
Compare the result
Assess whether the new approach genuinely improves time, quality, capability, or user experience.
Pro tip Compare against a defined current-state baseline.
Watch out Novelty can make an inferior workflow feel temporarily exciting.
- 6
Repeat and model the behavior
Continue identifying fixable friction and show others how proactive experimentation produces credible improvements.
Pro tip Share both successful and failed experiments to normalize learning.
Watch out Do not turn the model into a label that permanently categorizes people.
In the wild
An employee notices that weekly feedback is repeatedly collated by hand. Instead of accepting the process, they record the workflow, use an AI transformation skill to propose an assisted version, and test it on the next cycle.
→ Agency initiates the experiment while low tolerance directs effort toward meaningful friction.
A non-programmer discovers that AI now enables them to build a small internal tool. They prototype it, validate the output, and then apply the same proactive approach to other fixable workflows.
→ A new capability strengthens the person's expectation that deficient systems can be improved.
Common mistakes
Confusing agency with recklessness
Proactive experimentation still requires respect for security, authority, and irreversible consequences.
Turning low tolerance into negativity
The model calls for improving fixable systems, not constant criticism without action.
Rewarding prototypes without outcomes
Fast building matters only when the result creates a measurable improvement or useful learning.
Is it for you?
Best for
This is best for individuals and teams seeking a simple model for recognizing strong AI adopters and encouraging proactive improvement.
Not ideal for
It is not ideal as a substitute for governance in regulated, safety-critical, or irreversible work.
From the transcript
“The framework I've been using here, Kieran, is I think in a pre-AI world, the average person was more likely to be like lower agency,…”
“And I think all the people that we're seeing make this transformation are the exact opposite. They're super high agency. They're curious. They're out there…”
“And I suspect the best people you're working with fall into that like high agency, low tolerance bucket.”
From the episode
I Built a $20,000 AI Consultant You Can Have For Free