The Piano Is the Same. Who Creates the Music?

Imagine a concert hall.

Two pianists walk onto the stage. The same grand piano, the same tuning, the same score before them.

The first plays. Technically flawless. Not a single wrong note. The audience applauds.

Then the second sits down.

The same notes.

But this time, the audience holds its breath.

The difference was not in the piano. Nor in the score.

It was in what the person did with the instrument.

AI today is at a similar point.

A few years ago, access to a powerful model could itself be an advantage. Now roughly nine in ten organizations use AI in at least one business function, and most companies draw on similar large language models. The model is increasingly becoming an entry requirement rather than an advantage.

So the question is changing.

It is no longer ‘Do we have AI?’ that matters.

If everyone has access to the same models, why do some companies turn them into a real competitive advantage while others merely do things a little faster?

Companies that look alike are not the same

From the outside, two companies can look very similar.

They operate in the same sector. Sell to similar customers. Their CRM screens, sales stages, forecasting cycles, and management meetings are almost identical.

An article published in Harvard Business Review in February 2026 highlights precisely this point: when everyone has access to the same models, tools, and technology ecosystem, organizational context becomes the differentiator. Not the process written in the company’s procedure manuals, but how work actually gets done. Who watches which signals? What happens when an exception arises? In what order does a decision pass through whose hands? How does past experience shape today’s decision?

The model knows none of this on its own.

Two companies may look at the same map.

One knows the road.

The other can only read the map.

Repertoire: Not all data is an advantage

A pianist builds a repertoire over years. They have played the same piece hundreds of times, and each performance adds something to the last.

For companies, the equivalent is data.

But owning data does not create an advantage by itself.

What matters is data that competitors cannot easily buy and that becomes richer as the system is used: past transactions, customer behavior, collection patterns, pricing decisions, operational outcomes, and which decisions worked under which conditions…

McKinsey describes this as ‘privileged data’. If company-specific data enables better predictions, risk scores, or decisions, and each new use feeds more data back into the system, an advantage that is hard to imitate develops over time.

So the question is not how much data we have.

What memory do we possess that our competitors do not?

Interpretation: The model reads the score; context creates the music

Give the same score to two musicians and you will not hear the same thing.

In companies, the same data also means different things in different contexts.

A customer paying three days late may mean little on its own.

But if that customer’s order volume has fallen over the last six months, its credit limit is exhausted, and the sector is experiencing a cash squeeze, those same three days tell a different story.

The model can see the data.

But the picture remains incomplete until the company’s accumulated decision logic, exceptions, and commercial judgment are conveyed to it.

This is where HBR’s organizational context becomes visible: adding the knowledge embedded in how a company actually works to a standard model can produce different results from the same technology.

Rehearsal: The speed that matters is the speed of learning

When we discuss speed in AI, we usually think about how many seconds it took to perform a task.

I think a more valuable kind of speed is this:

How quickly does a company learn?

Choose a problem, test it on a small scale, observe the result, correct the mistakes, and scale it if it works.

McKinsey identifies the speed of learning and execution as a competitive moat in its own right in the AI era. One example is DBS Bank: within a broader transformation program, it reduced AI solution development and deployment time from 12–18 months to two or three months.

That speed does not come from buying a more powerful model.

It comes from changing how the organization works.

And then, the conductor

To me, this is the most interesting part.

In an orchestra, there is someone who does not play a single note: the conductor.

They determine who enters when, which sound comes forward, where to slow down, and where to stop.

In AI systems, the equivalent is governance.

Especially in areas such as finance, healthcare, and law, where errors are costly, producing the right result is not enough. Data sources, authority boundaries, control points, human approval, audit trails, and responsibility when something goes wrong are all part of the design.

McKinsey also argues that in high-risk domains, trust, together with explainability, auditability, data lineage, and human-in-the-loop controls, can become a direct strategic advantage.

I think there is an important distinction here.

The most dangerous orchestra is not the one that plays badly.

It is the one where everyone plays their own part perfectly, but nobody listens to anyone else.

Without direction, technical excellence only makes the noise more magnificent.

So where does the difference lie?

AI can read a contract.

Analyze financial statements.

Classify collection risks.

Forecast cash flow.

Prepare a presentation.

But your competitor can do all of this with the same model.

The lasting difference will therefore not be inside the model.

It will be in the system you build around it.

In the data you give it.

In how you carry the company’s memory into it.

In where you retain human judgment.

In which processes you redesign from the ground up.

In how quickly you can learn.

And in who notices when something goes wrong.

That is essentially the conclusion of McKinsey’s May 2026 study: using shared models is no longer enough. Competitive advantage comes from combining them with data, workflows, organizational structures, and capabilities that competitors cannot easily copy.

When the concert ends, nobody talks about how many keys the piano has.

What they remember is the performance.

I think AI is heading to the same place.

Soon, ‘Which model do you use?’ will not be a particularly interesting question.

The more important question will be:

If your model is the same as everyone else’s, what are you playing with it that nobody else can?

Sources

McKinsey & Company / QuantumBlack — May 15, 2026

From AI Table Stakes to AI Advantage: Building Competitive Moats.

Harvard Business Review — February 18, 2026

Rohan Narayana Murty & Ravi Kumar S., When Every Company Can Use the Same AI Models, Context Becomes a Competitive Advantage.