Toward AI-First: How AI Is Changing Business and What Comes Next

Some technologies enter our lives as useful inventions and only later reveal themselves as revolutions. Electricity, computers, the internet and mobile phones all opened opportunities that were far from obvious at the start. Over time, they changed how we live, work and do business. AI is now following a similar path.

The Scale of Change Is Rarely Clear at First

We tend to understand new technologies through the lens of what we already know. We use them within familiar processes, test different applications and gradually see opportunities we did not recognize at first. Over time, we begin to rethink the work itself and find new ways of creating value that would have been hard to imagine when we started. AI can advance quickly, while our understanding of how to use it struggles to keep pace.

The Internet Journey

We have seen this pattern before.

When the internet first arrived, people mostly used it to do familiar things in a new way. Companies built websites and moved information about themselves and their products online. In many cases, the website was little more than a digital shop window.

Banking followed the same path. Wells Fargo launched its first website in 1994. By 1995, customers could check their account balances online. Transfers and other banking services came later [1]. But the bank itself still worked much as before. It was, in essence, the same traditional bank with a new digital front door. The internet simply gave customers a new and more convenient way to deal with it.

As the internet spread, it started to change more of how banks and companies worked.

People began buying products online. Payments moved onto the web. Companies started working with customers, partners and employees in new ways. By the end of the 1990s, the internet was already being seen as a foundation for e-commerce and new ways of doing business [2].

And then came businesses that would have been difficult to imagine without the internet: marketplaces, social networks, streaming services, online banks and digital platforms.

The internet had started as a new channel. Over time, it became part of the infrastructure of modern business and everyday life.

The role of corporate IT has changed as well. In many large companies, IT had traditionally been a support function: the business defined the needs, and IT selected and delivered the technology to meet them. As technology became part of products, sales and customer service, that relationship started to change. Business and IT had to work much more closely together. Corporate management structures and operating models, however, changed far less. Most companies were still running largely as before.

Where We Are Today

For millions of people, AI is already part of everyday work and life. We use it to draft emails, prepare presentations, search for information, write code and solve everyday problems. At work, AI is moving beyond personal assistance. It can handle parts of end-to-end processes, work alongside people as AI agents and bring new capabilities into the business. In many cases, companies are still running in much the same way as before, while AI takes on a growing share of the work.

The first wave of enterprise AI adoption is already well under way. Employees are increasingly using corporate AI platforms that bring together capabilities from ChatGPT, Claude, Copilot and other AI solutions in a controlled enterprise environment. Organizations have launched chatbots, RAG solutions, document-processing tools, coding assistants and AI agents. Many have created AI Labs and centers of excellence, developed AI strategies, built corporate AI platforms and run dozens of PoCs and pilots.

In a sense, we already have the AI equivalent of the first corporate website.

The early results tell very different stories. Successful solutions have been moved into production and are delivering real business value. Others have been launched but are barely used. Some prove too expensive to take further. And several projects never make it beyond the pilot stage.

This is where the conversation begins to change. The early question was simple: where can we use AI in the business we already have? The next question goes much further: how should the business itself evolve around the new capabilities AI brings?

Why Many AI Projects Fall Short of Expectations

Banks and companies are still learning how to work with new technology, but many still approach AI using familiar management and project models. The challenge is that the way AI is introduced often shapes the results long before technology itself becomes the problem.

I would highlight five recurring reasons.

1. Leadership Understanding of AI

It is hard to ask the right questions before you understand what AI can realistically do. That is why the conversation often starts with broad requests: “We need GenAI,” “find us a use case,” or “another bank has done this, so we should too.”

The quality of the first questions matters. If the starting point is unclear, the projects that follow often lose direction.

2. Testing the Waters

When the potential value is still uncertain, management is naturally cautious about committing significant investment. The usual approach is to start small: appoint an AI advisor, hire one or two specialists, use existing IT resources and launch a few experiments.

What happens next can shape the entire journey. Strong early results can convince leadership that AI deserves broader investment. Weak or poorly selected pilots can leave the opposite impression: AI begins to look like just another IT system with an uncertain business case.

3. The Right Skills for the AI Lead and Team

Once management has a basic understanding of AI, the next challenge is hiring the right people with the right mix of skills. Too often, the focus is mainly on technical expertise: models, platforms, data and AI tools.

The AI Lead and team also need strong business and management skills: understanding the business, working effectively with business leaders, identifying the right opportunities, shaping joint initiatives and defining clear outcomes.

If these capabilities are overlooked, the team may end up delivering a technical result rather than a measurable business outcome.

4. Technology-Led vs Business-Led

There are two common ways to start. One begins with the business case and then identifies the technology needed to support it. The other starts with a platform, model or architecture and looks for suitable use cases afterward.

Both approaches can produce useful results. But when the starting point is the business opportunity and the expected outcome, the link to business value is usually much clearer.

A technology-led start can still produce solutions with a strong technical focus. The key questions then become especially important: What business case are we addressing, and what measurable value do we expect?

That value should be defined before the solution is built and measured once it is deployed.

5. Managing Risk as AI Scales

Early pilots are usually deliberately limited in scope. Teams choose relatively simple use cases, work with restricted data and focus on getting a visible result quickly. That makes experimentation easier, but it postpones the harder question: how do we manage AI risks once we want to scale?

At that point, the choice becomes more strategic. One is to stay conservative: rely mainly on internal models, restrict access to data and limit AI to a narrower set of use cases. The other is to strengthen AI risk management and security so that more advanced applications can be used safely.

Scaling therefore requires more than access to a wider range of AI models and complex AI platforms. It requires a more mature approach to security, data access, compliance, accountability and controls, designed for the realities of enterprise AI.

As experience grows, a more fundamental question emerges. How far can we go by adding AI to a traditional business model, and when do we need to reshape the business itself around the new capabilities enabled by AI?

From Experimentation to AI-First

For the past 10 to 15 years, much of the corporate agenda has been shaped by digital transformation. Companies digitized existing processes, created new channels and moved more services online.

AI goes deeper. People are sharing more work with AI, delegating parts of processes to agents and finding new ways to combine human judgment with AI capabilities. This new division of work is changing roles, responsibilities, decision-making and how teams work.

We can already see early signs of this shift. At OpenAI, Codex started as an AI tool for developers and is now also used across functions such as HR, Legal and Finance [7]. Danske Bank reports that around 16,000 employees use internal GenAI solutions each month and has publicly stated its ambition to become an AI-first bank [8].

PwC, Deloitte and EY are also developing managed-service models where people, AI agents and automation jointly run entire business functions for clients. Their offerings now span areas such as IT operations, finance, procurement, cyber, tax and compliance [9–11].

The same pattern is appearing across the market. Technology providers are moving beyond standalone tools, companies are embedding AI into day-to-day work, and consulting firms are changing how services are delivered. AI is becoming part of how the business itself operates.

What AI-First Could Look Like

The contours of an AI-first business are taking shape. Data and AI are moving closer to the core of how the business operates, while the full model is still evolving.

BCG and McKinsey both focus on redesigning work and the organization. BCG looks at how tasks are divided between people, AI assistants and agents, while McKinsey emphasizes broader choices around decision-making, organizational capabilities and value creation [3,4].

Deloitte and Microsoft place more emphasis on how people and AI work together at scale. Deloitte focuses on orchestration, governance and accountability. Microsoft’s Frontier Firm concept puts human direction and judgment at the center while agents take on more execution [5,6].

The terminology differs, but the direction is similar. People focus more on judgment, direction and accountability, while AI takes on a larger share of execution. This new division of work begins to reshape roles, decision-making, governance and business processes.

What's Next?

We are still learning what an AI-first bank or company may look like in five or ten years. Its broader impact on everyday life, work, communication and decision-making is still unfolding. But one thing is already clear: as people change how they work and what they expect, businesses will have to change with them.

For leaders, the priority is to move from broad ambition to a more practical set of questions:

• Which business areas or operations can AI materially improve today?
• What measurable business value do we expect?
• What needs to change in processes, data and the way people work?
• What capabilities should we build now to move from isolated AI initiatives toward AI-first?

These questions help keep the focus on business value while building the foundations for scale. They also create a bridge between today’s pilots and the broader changes that may follow.

AI is still at an early stage of this transition. The internet first changed how people communicated, searched, bought and worked, and over time it reshaped entire industries and business models. AI may follow a similar path, potentially reaching even deeper into how work is done and how companies are run.

The companies that start building understanding, capabilities and governance now will be better placed to shape what comes next, rather than simply react to it.

References

[1] Wells Fargo. First in Online Banking. Source

[2] Microsoft. Microsoft Announces Comprehensive Strategy to Accelerate Participation of Businesses and Consumers in E-Commerce. March 4, 1999. Source

[3] BCG. Design Your Company for AI, Not AI for Your Company. April 23, 2026. Source

[4] McKinsey & Company. The Key to AI Value Is Hiding in Plain Sight: Your Operating Model. September 9, 2026. Source

[5] Microsoft. 2026 Work Trend Index Annual Report: Agents, Human Agency, and the Opportunity for Every Organization. May 5, 2026. Source

[6] Deloitte. Rewiring the Enterprise Operating Model for AI Scale. June 29, 2026. Source

[7] OpenAI. How Agents Are Transforming Work. June 25, 2026. Source

[8] Danske Bank. Danske Bank further strengthens Generative AI focus with new leadership appointment. June 16, 2025. Source

[9] AI-enabled Managed Services: PwC. Accelerating Outcomes: Inside PwC’s AI-Enabled Managed Services. June 25, 2026. Source

[10] Deloitte. Operate Managed Services. Source

[11] EY. EY-Microsoft Alliance for financial organizations — Managed Services. Source