Why not just build it? What could make us say no?
Another AI idea enters the Monday meeting.
Someone has tried a new tool. Someone else saw a competitor's demo. A team has found a task that looks as if it could be automated.
Within ten minutes, the idea has an owner and a prototype deadline.
Now, this meeting feels productive.
But nobody has asked whether the problem is important enough to solve, what evidence supports the team's confidence, or what would make them stop.
This is how internal AI slop begins: plausible pilots, assistants, and automations that were easy to start but never had to prove they deserved investment.
I think organizations today do not have an AI idea problem anymore. They have a decision problem.
If you are an AI champion, AI adoption lead, or AI product manager, two questions can immediately improve the next conversation—and help you become more strategic in how your organization chooses what to build.
Meetings make ideas move
Most meetings are designed to exchange information, resolve immediate issues, and assign next steps.
They are rarely designed to compare AI opportunities, expose assumptions, connect an idea to a strategic priority, or decide what should not move forward.
That changes the questions people ask.
Instead of:
- Is this problem important enough to solve?
- What evidence supports our confidence?
- What would make us stop?
The meeting moves toward:
- Who owns it?
- How quickly can we prototype it?
- When can leadership see a demo?
The problem is not the meeting itself. The problem is the absence of a decision process inside it.
If you are the person responsible for AI adoption, you can improve that conversation with two questions.
Question one
Ask the question everyone in the room is already thinking:
"Why not just build it?"
Then let the group make the strongest possible case for moving immediately.
They will tell you that building is cheap. That competitors are moving. That the organization needs momentum. That a prototype will teach them more than another discussion. That they understand what employees or customers need. That there is no time to pause.
Those answers reveal the pressure behind the decision.
That pressure matters.
Pressure that stays unspoken does not disappear. It steers the decision from underneath. Once it is on the table, the team can examine it rather than being moved by it without noticing.
The first question gives the pressure somewhere to go.
It still does not tell you whether the idea is worth building.
Building has become cheap. Commitment has not.
The cost of producing a plausible AI prototype has fallen dramatically.
A small team can turn an idea into something convincing in a few hours. They may not need an engineering team, a formal budget, or leadership approval to begin.
That is real progress.
But a cheap prototype can still create an expensive commitment.
The moment something exists, the conversation changes. The organization has something tangible to defend. Stakeholders become attached. More time gets allocated. More meetings. More people are involved. Integration, governance, data, adoption, maintenance, and ownership are now getting prioritized.
The prototype was cheap. Everything it set in motion was not.
When building itself created friction, teams had to make choices before they began. They had to justify the budget, secure technical capacity, and persuade someone that the opportunity deserved attention.
AI removes much of that early friction. But removing friction creates another problem: when nothing forces a choice, almost everything begins to look worth building. General Magic offers a cautionary example.
The company that could build anything, built everything
General Magic is an extreme example of what happens when the ability to build stops acting as a filter.
In 1990, Marc Porat spun a project out of Apple with John Sculley's support. Its founding team included Andy Hertzfeld and Joanna Hoffman, two of the people who helped build the Macintosh.
It became one of the most desirable places to work in Silicon Valley.
The engineers had deep funding, exceptional talent, and extraordinary freedom. There was no strong product-management function continuously ruling ideas out of scope. If the team believed something was worth building, it could build it.
So it built a touchscreen, network protocols, software agents, downloadable applications, animated emoji, and its own version of technologies that would later become commonplace.
Many of the ideas were remarkable.
Together, they became a product that was late, expensive, difficult to understand, and dependent on an infrastructure bet that the open internet was already overtaking. The Sony Magic Link launched in 1994 and sold only a few thousand units.
Timing was part of the failure, but it was not the only reason.
Another way to read the story is that the team had the freedom to pursue almost every idea and no strong mechanism for deciding which ideas should survive.
Freedom does not create focus. Freedom is not a strategy.
General Magic was building one product with too many features. Your organization may be accumulating separate AI pilots, assistants, and automations. The scale is different, but the mechanism carries across: when nothing rules an idea out, every attractive idea survives. The cost appears later as AI slop.
AI has put ordinary teams in a smaller version of those conditions. Almost every idea is cheap enough to explore.
A company with no explicit reason to say no will keep finding reasons to say yes.
That is why you need the second question.
Question two
Once the group has made its case for moving quickly, change the direction of the conversation.
Ask:
"What could make us say no to this idea?"
This question examines the team's confidence in the idea.
It asks the team to notice where that confidence weakens and what could change its mind.
Perhaps the customer or employee problem is less urgent than the team assumes. Perhaps the idea supports no meaningful strategic priority. The necessary data may be unreliable, inaccessible, or restricted. The workflow may be poorly understood. The people expected to use the solution may have little reason to trust it. A simpler intervention may solve the problem without AI. Or the team may realize that it cannot name the evidence that would cause it to stop.
Personal stakes will surface too. Someone may have spent weeks on a pilot that nobody uses. Someone else may fear being seen as slow. Another person may be accountable if the initiative fails.
These answers do not test the idea.
They expose what the group's confidence rests on, where evidence is missing, and which conditions would justify a pause or a no.
A question can expose an assumption. Only evidence can test it.
Two questions make the meeting better
The questions do different work:
- Why not just build it? reveals the pressure pushing the group toward action.
- What could make us say no to this idea? reveals the strength and limits of the group's confidence.
Before the questions, the group has an idea, urgency, and the feeling that the next step is obvious.
After the questions, it can see:
- What is pushing it to move quickly
- Which beliefs are being treated as facts
- Where its confidence begins to weaken
- What evidence it still needs
- What could cause it to build, pause, or stop
That is already a much better meeting.
The group may still decide to build. It may pause while it investigates a critical assumption. It may decide that the opportunity does not deserve further attention.
The questions cannot guarantee the right decision.
They make the reasoning visible enough to improve it.
But when AI ideas appear every week, two questions inside a meeting are only the beginning. The organization needs a repeatable place to make these decisions.
When the ideas keep coming, change the format
The more strategic move is to stop treating AI opportunities as ordinary agenda items.
Give the decision its own structured space.
An AI Problem Framing workshop is a full-day decision-making process for a cross-functional team. It takes the ideas already circulating through the organization and evaluates them against the same questions:
- Does this solve a problem the business has decided matters?
- Who experiences that problem, and how urgent is it?
- How does the work happen today?
- What measurable value would the idea create if it succeeded?
- What assumptions, constraints, and data dependencies could weaken it?
- What evidence is needed before further commitment?
This is not another brainstorming session. The organization already has enough ideas. The purpose is to decide which ideas should move forward, which need more evidence, and which should go no further.
How ALEC used one day to decide before building
ALEC Holdings, one of the Gulf's largest construction companies, had already moved well beyond basic AI awareness.
Thousands of employees were using approved AI tools. The company had a custom AI agent in production and an established Innovation Champions network.
Its problem was not generating enthusiasm.
Across finance, procurement, design, project management, HSE, and site operations, the organization had collected more than 45 credible AI opportunities.
The challenge was deciding which ones should move first.
In Dubai, ALEC brought 55 people into six cross-functional AI Discovery Pods. The room included domain experts, senior sponsors, Innovation Champions, and IT.
For one day, the teams ran AI Problem Framing.
Every opportunity went through the same structured evaluation. The teams examined the underlying business problem, the current workflow, the expected value, the strategic connection, and the reasons an idea might not deserve investment.
By the end of the day, six teams had converged on six validated AI use cases.
Only then did those use cases move into AI Workflow Sprints, where the teams redesigned the work and created AI agent blueprints.
The framing day did not slow ALEC down. Only those six opportunities moved into the next phase. The remaining ideas did not become projects simply because someone had proposed them in a meeting.
That is the difference between collecting AI ideas and building an AI decision-making capability.
This is how you become more strategic
If you are an AI champion, AI adoption lead, or AI product manager, your value is not measured by how many ideas you help activate.
It is also measured by the quality of the filter you create.
You become more strategic when you help the organization distinguish:
- Urgency from evidence
- Confidence from knowledge
- A cheap prototype from an expensive commitment
- An interesting idea from an opportunity worth investment
You do not need to decide for the group. You need to improve the way the group decides.
Start in the next meeting with two questions:
Why not just build it?
What could make us say no to this idea?
Those two questions will leave the meeting in a better place than it started.
And when the flow of ideas becomes too important for an ordinary meeting, change the format. Bring the right people together for a structured day and decide what is worth building before the organization starts building everything. That is how you move from supporting AI activity to shaping AI investment.
