How to Implement AI and Scale AI Projects Successfully

News of high enterprise AI adoption has been making headlines, but in reality, how much of it is actually translating into greater business performance?

An MIT report reveals that despite $30–40 billion in enterprise investment in GenAI, 95% of organisations are getting zero return. In fact, only 5% of integrated AI pilots are delivering ROI, and the vast majority are stuck with no measurable P&L impact.

In my interactions with businesses over the past year as Senior Manager, Automation and AI at Org, I have observed a similar pattern.

From organisations wanting to implement AI to those looking to move beyond experimentation and achieve meaningful transformation, the roadblocks are often the same. And in most cases, the issue is not the AI tool or its quality, but the approach to AI adoption.

In this article, I share my insights on how to implement AI and scale it successfully, so that your organisation can see the strongest returns from AI implementation.

Here’s my step-by-step guide on how to do it:

1. Address the Problem to Solve First

A common mistake many organisations make is treating AI as a silver bullet that can solve any challenge without first understanding their problem in detail. If the problem is not defined, then it will be difficult for any solution, even one powered by AI, to address it effectively.

One reason people have this perception of AI is that their personal experience of it is usually limited to point solutions that help them create a holiday itinerary or summarise a document.

In those cases, AI works efficiently and quickly because the task is specific and the dataset is narrow. But when it comes to implementing AI in enterprise environments, it requires far more groundwork.

Multi-step, complex business processes require flows, guardrails and auditing. If you have not defined those workflows before bringing an AI solution into the business, then it is going to be very difficult to get results. Imagine trying to fix a plane while you are flying it; that’s what you would effectively be trying to do.

The commonly seen approach of purchasing an AI tool first and then expecting it to make things better is one of the main reasons most AI pilots fail. The right way to implement AI is to first understand the problem you want to solve, then define the scope, and only then bring in an AI solution to address it.

2. Have a Solid Transformation Strategy

When an organisation decides to implement AI, the decision is often made at the executive level with a focus on business benefits such as improving performance, reducing costs or saving time.

But senior leaders may not have carried out an enterprise transformation assessment to understand whether the organisation is ready for the change they are proposing.

That matters because employees often see the initiative from a completely different lens. Their primary concern may be whether their job is going to be discontinued or whether they are going to be made redundant. Without a strong transformation strategy, it becomes difficult to bring employees with you on the journey.

The way I explain it to organisations is this: nobody gets to Friday afternoon at 2 pm and says, “Okay, well, I’ve done everything I need to do. I can finish working early.” There is always something else to do.

What AI does is take away the time-consuming, manual, repetitive work and allow people to do the work they would like to get to on a Friday afternoon, but are often too busy doing manual tasks to reach.

That is why the transformation strategy matters. It is not just about business efficiency. It is also about showing employees how AI will change their work in a way that creates more opportunity.

3. Address the Human Side

While having a transformation strategy is paramount, it is equally important that the strategy is human and accessible.

For employees working in different departments, such as finance, HR, or marketing, the company’s strategy should be translated into terms that make sense to them. We often see companies take this lightly, but it needs to be made much more evident in most organisations.

Especially with the fear that AI will replace them, employees tend to be afraid of change. Senior management needs to take that human nature into account and actively allay those fears. Explain how implementing AI will free employees up to do more strategic work, give them more opportunities and help them progress in their careers.

The human side of your strategy also means recognising that although a strategy should be defined at the outset, it should not be locked in stone. As the organisation transforms, the strategy may need to evolve to reflect the benefits already realised through AI, as well as any additional opportunities that emerge.

4. Don’t Treat Pilots as the End Goal

Many organisations succeed with AI pilots, but when it comes to broad deployment, they run into operational challenges. I have noticed a few reasons why this happens:

• Often, an AI pilot project is run by a dedicated team, and they have a clearly defined goal and objective to achieve.
• While the ultimate solution would require integration with different tools, organisations sometimes simulate the integration during the pilot because it is more achievable in a shorter timeframe. This narrows the delivery focus, but it can also introduce gaps if the path to production has not been thought through.
• Since pilots work on a narrow focus, aspects such as security, legal requirements, governance and different jurisdictions may not be considered in detail, which can create problems later when moving to production.
• During the pilot, organisations may engage subject matter experts intensively so they can help the project succeed. But when the pilot moves into production, those same people may need to repeat the work a second time and quite rightly ask, “Didn’t I tell you this already?”
• The subject matter experts may have been seconded to the pilot and were able to focus on it intensely, but when the solution moves to production, they may no longer be seconded and instead have to carry out the migration work alongside their day jobs, which can be more difficult to achieve.

I always tell businesses to remember that an AI pilot is not the end goal. The pilot is there to justify the final spend, validate the proof of concept, and then move forward into production.

5. Focus on Outcomes Rather Than Specific Tools

My advice to businesses that are struggling to assess AI tools is to think carefully about the sequencing of your vendor selection. Start with the outcome you want to achieve rather than focusing on the AI tools. Otherwise, you can end up in a “hammer and nail” situation, where you force-fit a specific vendor’s software (such as Microsoft or Google) to your problem regardless of whether it is the right fit.

At Org, what we do when we are talking to clients is we understand the outcome they want and then determine how to technically best deliver it. Often, companies come to us with two related concerns: they want to use AI but do not know where to start, or they have already decided to start implementing AI but do not know how to begin.

Our approach is to run an ‘Idea to Innovate’ workshop with a group of people from that organisation. That may include representatives from the executive level, finance, HR, manufacturing and sales.

We facilitate this workshop where they discuss their problems in terms that make sense to them. The result is that people across different parts of the organisation have discussed their issues and defined them in terms that are meaningful to the business.

For example, it is not me telling them, “Your finance problem is the biggest problem.” It is them saying, “Our finance problem is our biggest problem because we have regulatory issues, payment issues and credit issues.”

Once that is clear, they have a list of priorities they can focus on. We also help them understand how to get started. Based on the outcome they have identified, we suggest a tool, or sometimes a combination of tools, that can help them achieve it.

6. Recruit Internal AI Advocates

If you want your pilot to scale, one thing you have to do right at the beginning is find internal AI advocates.

Pick a friendly face, somebody in your organisation who is enthusiastic about AI or potentially has some experience with it. Start the pilot with their department. They will become a natural advocate for the adoption of AI within the organisation.

People need to be brought on the journey to make sure that what you are delivering is a success. You can only go as fast as the slowest individual in your group, so you need to take them with you.

7. Think Beyond Point Solutions

An AI solution intended for enterprise scale must go beyond point solutions. It needs to be auditable. It needs logging. It needs to be integrated into your IT infrastructure. It also needs to align with your security considerations.

This matters especially in high-volume environments. If an organisation is processing 10,000 or 20,000 invoices every month, it needs to be certain that the solution will perform reliably every time and handle every exception the business throws at it.

8. Get Guidance from AI Experts

Choosing the right AI implementation partner can make a significant difference to the success of any project.

At Org, we have experience delivering AI solutions across a range of industries, business functions and jurisdictions. For example, we recently worked on a project involving Chinese medical records in Hong Kong, where AI was used to translate records, reconcile them against insurance thresholds and support payment recommendations.

We have also worked with companies in the Netherlands where AI is being used to manage car accident claims, assessing documents such as police reports, damage reports and hospital bills to support underwriting decisions.

We are also currently working with the National Transport Authority of Ireland (NTA) on developing an AI governance framework that defines how AI will be used across the organisation.

This means we can support organisations at every stage of the journey, from AI governance and strategy, to project scoping through Idea to Innovate workshops, to implementation. Our experience means we can help organisations not only define their priorities, but also turn them into practical, scalable solutions.

If your organisation is exploring how to implement AI more effectively, speak to our team to learn how we can help you define the right strategy and build an AI solution that scales.