What is 'deployment'?
In the context of AI transformation, deployment means slotting AI into the gaps of your existing workflow while leaving the sequence of work untouched. You break your current process into tasks and drop AI in between them to gain speed. The way of working stays the same; only the tool changes.
This pattern is not new. When the iPhone arrived and mobile was the topic of the day, many companies created a CDO (Chief Digital Officer) role and drove digital transformation with that person at the center. The AI of the past two or three years looked similar: appoint an AI executive (a CAIO, Chief AI Officer) and say 'we're working hard on AI too.' Up to that point, almost every company was the same.
The problem is that this only makes the hands and feet faster. What used to take six months now takes a little less, but the way the company actually wins in the market does not change. AI was added, yet the competitive board stayed the same.
How is 'reshape' different?
Reshape means re-forming — not inserting AI into the work, but tearing up the sequence of the work itself. BCG places the core of AI transformation here. With the very same AI, deployment and reshape produce completely different results.
Take new product planning. Developing and planning a single new product at a consumer-goods company usually takes six months to a year. With deployment, you set up the existing development steps as tasks and insert AI between them to make each step a little faster. The sequence is unchanged.
Reshape rewrites the sequence. The old way pours six months to a year into a single opportunity angle — one hypothesis about which product to sell to which customer. With reshape, AI helps analyze the market and identify opportunity areas, rapidly generating many opportunity angles at once rather than just one. You then narrow toward the one with the highest odds of success. Done this way, six months to a year shrinks to one or two.
The point is not 'we did it faster' but 'we changed the order.' A process that clung to one hypothesis for a long time is reconstructed into one that opens many hypotheses at once and narrows them down. AI was not plugged into the old process; the skeleton of new-product planning was rebuilt.
What is Enterprise IQ?
But what actually makes this redesign possible is not the tool called AI. BCG calls that point the company's brain — Enterprise IQ. It is your own logic for how you observe the market and where you draw insight (the opportunities and understanding others fail to see) from.
An analogy: AI gives you hands and feet that are fast, and many of them. But the real worth of those hands and feet lies not in their speed or number, but in whether they move together with a brain that has been implanted with your strategy for winning in your market. Hands and feet made fast without a brain are just busy without direction. Success or failure turns on how well you have cultivated this Enterprise IQ.
Crucially, no outsider can build this brain for you. Neither OpenAI nor Palantir (a large data-analytics company) can create this part. It belongs to the realm of internalization — only the company that understands its own logic can build it. Buying an AI model or solution does not conjure it into being.
So the order of the questions changes. Not 'where do we use AI,' but 'what makes a good new product succeed' and 'what have we been winning with.' AI is merely the hands and feet that put that winning logic into action.
Why do most fail — the illusion of 'working hard'
Many companies say they understand AI, but few approach it in a way that produces results. The most common trap is that an 'AI transformation' becomes absorbed not in the transformation but in adopting AI technology itself. AI is a means, yet it quietly becomes the end.
This is where the illusion of 'working hard' appears. When you insert AI here and there through deployment, the company looks, on the surface, as if it is handling AI actively and well. But over time, people start saying: 'We ran AI for more than a year and spent a lot of money, but we don't know what actually changed.' It looked busy, but the board stayed the same.
Goldman Sachs is a telling case. CEO David Solomon said that before 2024 the firm drove enterprise-wide AI transformation by running more than a hundred POCs at once. A POC (Proof of Concept) is a small trial to see whether something works. But the firm failed to produce real successes, and after acknowledging that failure, from 2024 it made a major pivot to a philosophy of focusing on the five or six most important things that address the fundamentals. The lesson: focusing on the few that change the roots beats launching a hundred.
The real hard part is people, not technology — 70% of AX
People tend to assume that what makes AI transformation hard is the technology — the models, the data, the infrastructure. But in the field, the harder part is the organizations and people leading it, and change management (managing the shift in how an organization works and behaves). BCG estimates that about 70% of the difficulty in AI transformation comes down to people.
So more weight has to go on how it is applied in the field and how people will change than on advising AI strategy and designing platforms and development. No matter how good the AI is, results will not come unless actual practices, processes, and — where necessary — organizational structure change alongside it. If you redesign under the banner of adopting new technology while people's way of working stays the same, the transformation is only half done.
So what should you do — top-down and the opening for Korean companies
First, a top-down approach — top management pulling it directly — matters more and more. Compare a company where the CEO personally orchestrates AI and invests heavily, one that treats it as mid-priority, and one that does not: business results diverge starkly. Winners and losers become clear. BCG says that pushing from the top with the resolve to overcome the difficulties is, in effect, the only way to guarantee success.
Here Korean and Asian companies have a special opening. In the AI era, there is almost nowhere that is genuinely ahead. It is a relatively new technology that everyone is grappling with together. Korean companies are unusually receptive and proactive toward this kind of change. Marketing, for instance — the core of consumer goods — used to be a game hard to win when Western giants with enormous capital pushed in. But if you can execute marketing at far lower cost through the power of AI, that board is worth flipping cheaply.
Large-scale industries Korea does well in — autos and shipbuilding, and lately cosmetics and food — are areas that can benefit greatly from AI. It is true that the U.S. leads in foundation models (base AI models like GPT and Claude) and core technology, but it is not yet clear whether the U.S. is also the most advanced at actually using them. The game of application is only beginning, and which country leads it is not yet decided.
A starting point for practitioners
The first thing to do is to put into words the way you have been winning in your market. If that winning logic lives only in a few people's heads rather than in a document, there is no way to embed it into AI. Ask first: 'Whose head is our company's winning logic living in right now?'
Next, do not start by listing where to insert AI. Pick one of your core functions — new-product planning, sales, or marketing — and redesign the sequence of that work from a blank page. Ask, 'If AI can generate many angles at once, how does the order of this work change?' and the thread of reshape appears.
Finally, do not launch a hundred things at once; focus on the five or six that touch the fundamentals. And from the very start, design the change in the people and the organization doing the work, not only the technology. No matter how fast you build the hands and feet, if there is no brain to embed, all you are left with is time spent and money gone without knowing what changed.