Before You Ask AI What to Do, Tell It What Matters

September 8, 2026 · By Michel Valenzuela, Founder of Daryl

Two people can face the same opportunity and reasonably make different decisions.

One accepts a demanding project because the income matters immediately. The other declines because protecting an existing commitment matters more.

The opportunity is the same. The priorities are different.

That distinction is easy to lose when asking AI for advice.

“Which option is best?” sounds complete. But the answer depends on what you are trying to achieve, what you need to protect, and what you can afford to give up.

Before asking for a recommendation, give the conversation those foundations.

Make the objective specific enough to guide you

“I want to grow my business” is a direction. It leaves many decisions unresolved.

Growth could mean more revenue, better margins, a larger team, stronger client relationships, or more time to develop your own products.

Those goals can lead to different choices.

A project that increases revenue may reduce the time available for building something you own. Hiring may create capacity while adding responsibilities you do not want yet.

Try explaining the objective in terms that can guide the decision:

“Over the next six months, I want to maintain reliable client income while protecting two days a week for product development.”

Now the assistant has something meaningful to work with.

You do, too.

State what cannot move

Some parts of a decision are open to negotiation. Others are commitments.

A promised delivery date, a fixed budget, or a limited number of available hours can determine whether an option is workable.

Put those constraints into the conversation early.

For example:

“I have twelve hours a week available. I cannot move the deadline for my existing client, and I cannot add another recurring commitment this month.”

An option that ignores those facts may sound attractive while being impossible to carry out.

Separating firm constraints from preferences also reveals where flexibility exists. A preferred start date might move. A signed commitment may not.

That distinction helps the discussion stay practical.

Explain the cost you are prepared to accept

Every worthwhile option can ask something of you.

More income may require more time. A faster launch may require a smaller scope. A new direction may mean letting go of work that has become familiar.

The question is which cost you are willing to accept for the outcome you want.

You might say:

“I’m willing to reduce the initial feature set to meet the launch date. I’m not willing to release an experience I cannot support reliably.”

That statement gives the comparison a boundary.

It also helps you recognise when a proposed solution achieves the objective by sacrificing something you intended to preserve.

Include priorities that are difficult to quantify

Some of the most consequential criteria will never fit neatly into a spreadsheet.

You may want work that develops your judgment. You may value independence. You may care about having enough attention left for your family or your own creative practice.

Those considerations belong in the decision.

If you leave them out, the discussion may favour whatever is easiest to measure: fee, duration, output, or apparent efficiency.

An assistant can only work with the priorities you make available. Tell it when a less measurable consideration should carry weight.

“The fee matters, but I also want to avoid another project that leaves me too exhausted to work on my own ideas.”

That is relevant information.

Check whether your priorities conflict

Once you state what matters, you may discover that you are asking for incompatible outcomes.

You want to accept every promising opportunity and reduce your workload. You want to launch quickly and include every planned feature. You want more freedom while maintaining every existing obligation.

An AI conversation can help you examine those tensions.

Ask:

“Which of my stated priorities conflict? Where will I need to choose or change the arrangement?”

Sometimes the conflict can be reduced. A smaller engagement may preserve both income and time. A staged release may support both momentum and quality.

Sometimes you need to decide which priority comes first.

Making that choice explicit is more useful than allowing it to happen through overcommitment.

Ask the assistant to use your criteria visibly

After explaining the objective, constraints, and acceptable trade-offs, ask for a comparison you can inspect.

“Compare these options using the priorities I’ve described. Explain where each option fits, where it conflicts, and which assumptions need checking.”

Then review whether the assistant actually followed your criteria.

Did it overlook a firm commitment? Treat a preference as mandatory? Introduce an objective you never stated?

Correct the comparison before relying on it.

A useful answer should make the relationship between your priorities and its recommendation understandable.

Keep room to revise the brief

Explaining what matters can change your understanding of the decision.

You may realise that revenue is less urgent than you thought, or that a deadline has become important mainly because you announced it too early.

You are allowed to revise the criteria.

What matters is making the change consciously and recognising its consequences. A new priority should lead to a fresh comparison, rather than being quietly added to justify an answer you already wanted.

The conversation can help you notice that shift.

Why this matters for Daryl

Daryl is a personal AI for Mac, currently in development, designed around local conversation and support for decisions.

I want it to help people bring their priorities into view and examine options in relation to their actual circumstances.

That begins with the person using it.

An assistant can help organise a comparison, identify a conflict, or ask about a missing constraint. Your goals give that work its direction.

The experience I’m building toward should make those connections easier to understand.

A recommendation becomes useful when you can see why it fits what matters to you.

Before asking AI what to do, tell it what you are trying to make possible.