THE SKATING CEO · THINK WITH AI 05 / 07

ISSUE NO. 05

Teach AI How
You Decide.

Do not just tell AI what you want.

Give AI the principles behind your decisions. Your rules will be different from mine, but they should be explicit.

ADAM JUKES FOUNDER, ONE BLADES

QUICK VERSION

READ THIS IF YOU ONLY HAVE TWO MINUTES

05

Most people tell AI
what they want.
Very few tell it how they judge whether an idea is good.

At ONE Blades, major decisions can be tested against five questions:

01

Does this benefit skaters, coaches, parents, or creators?

02

Does this strengthen inline figure skating?

03

Does this protect creator economics?

04

Does this respect the phase model?

05

Could this decision be explained publicly without spin?

That is far more useful than simply asking:

Is this a good idea?

The AI can now evaluate the idea against the rules of the actual company.

Your framework will be different.

Create three to five questions that reflect what really matters in your decisions.

Also tell AI:

  • What has already been rejected.
  • What is non-negotiable.
  • What evidence would justify reopening a decision.

Good personalization is not only
“know me.”

It is
“understand how I decide.”

THE WHOLE IDEA

Teach AI the criteria behind your judgment.

Then it can evaluate ideas against what actually matters to you.

END OF QUICK VERSION CONTINUE FOR THE FULL PAPER ↓

GO DEEPER

The difference
is often criteria.

The same opportunity can be good for one person and terrible for another.

The difference is often not information.

It is criteria.

01

A real
ONE example.

Imagine a manufacturer becomes excited about a future product idea and wants ONE to move faster.

Generic business logic might say:

There is momentum.
Capitalize on it.
Move now.

But ONE has a phase model.

Phase 1 remains focused on EDGE and inline skating.

Future products do not become public simply because someone else is enthusiastic.

Manufacturer enthusiasm
is information,
not direction.

The AI needs to know that.

Otherwise, it may repeatedly recommend decisions that look commercially exciting but create mission drift.

02

The ONE
Five-Question Test.

ONE uses five questions for major decisions.

They are powerful because they convert broad principles into something testable.

01

Does this benefit skaters, coaches, parents, or creators?

This protects Skater First.

02

Does this strengthen inline figure skating?

This protects the mission.

03

Does this protect creator economics?

This prevents growth ideas from quietly weakening the creator model.

04

Does this respect the phase model?

This stops distraction masquerading as opportunity.

05

Could this decision be explained publicly without spin?

This protects honesty.

Now the AI has a filter.

03

Build your own
decision test.

Do not copy ONE's questions unless they genuinely fit your life.

Find your own.

A FREELANCER

  1. Does this work pay enough to justify the time?
  2. Does it strengthen the portfolio I want?
  3. Do I respect the client?
  4. Will it create recurring stress?
  5. Would I still want it if nobody else knew I got the job?

A PARENT CONSIDERING A MAJOR PURCHASE

  1. Can we comfortably afford it?
  2. Does it solve a real problem?
  3. Will we still use it in a year?
  4. What are we giving up to pay for it?
  5. Are we buying it because we need it or because we are excited today?
04

Maintain a
retired ideas list.

AI is very good at generating common strategies.

That means old ideas can keep resurfacing.

At ONE, several systems have been deliberately retired.

If those decisions are not documented, the AI may innocently suggest them again.

A retired ideas section
saves time.

It says:

RETIRED IDEA

We considered this.

We rejected it.

Here is the reason.

Do not reintroduce it unless something material has changed.

This does not mean every old decision is permanent.

It means an idea should not keep returning simply because the AI does not know it has already been considered.

Remember what you rejected,
not only what you chose.

05

Ask AI to challenge
your own framework.

Decision rules should not become religion.

Occasionally ask:

01

Which of these rules might now be outdated?

02

What situation would justify an exception?

03

Are two of these principles in conflict?

That keeps the framework alive.

The rules should guide
your judgment.

They should not replace it.

06

Use AI
as a red team.

Once the AI understands your rules, ask it to attack the idea.

For example:

Evaluate this proposal against the five questions.

Then give me the strongest argument for doing it and the strongest argument against it.

Identify any assumption that is carrying too much weight.

That is much stronger than asking for generic pros and cons.

07

Teach it what matters
before asking it what to do.

AI becomes more useful when it understands not only the destination, but the rules of the road.

Teach it what matters
before asking it
what to do.

That does not mean AI should agree with you.

It means it has enough context to challenge the actual decision you are making, rather than giving you generic advice for somebody else's situation.

THINK WITH AI · ISSUE NO. 05

Good personalization
is not only know me.

It is understand how I decide.

NEXT ISSUE 06 / 07

ISSUE NO. 06

Turn Repeated Conversations
Into Systems.

WRITTEN BY

Adam Jukes

Founder of ONE Blades · The Skating CEO

Professional skater, coach, performer and founder building ONE Blades and documenting the work, decisions and ideas behind it.

Published
Last reviewed and updated