Expert roundtable: practical advice on how to successfully roll out AI at work

Article

AI adoption is rapidly transforming the workplace. Nearly three-quarters (72%) of companies are encouraging their employees to explore Gen AI tools for routine tasks and enhanced productivity, and 42% of employees are already using the tech in their daily work. Although the tool is revolutionising how some businesses operate, rushing to implement it for implementation's sake could be a costly error — a systematic, pragmatic approach is needed.

On Tuesday, November 12th, Robert Half held a breakfast roundtable with guest speaker Harrison Jardine, business process improvement and intelligent automation lead at our sister company Protiviti, to discuss the exploration of AI for sustainable advantage.

Over the course of an hour, our roundtable attendees discuss practical AI applications, realistic digital transformation strategies, and how to begin exploring AI adoption in tandem with effective upskilling and change management.

The most successful AI pilot rollouts start with a solid business case or challenge and clear success metrics. "Please don't be led by the technology," says Harrison. "As with any other technology that has come before, AI is just another capability. It's ultimately about first understanding your biggest business challenges and opportunity areas before trying to work out which technology fits."

In real-world terms, AI can be deployed for manual tasks that don’t necessarily require a human touch. For example, communications mining to prioritise customer inquiries and dedicate staff resources to moments that matter.

'We implemented this at a global vehicle leasing firm," says Harrison. "We managed to move their first touchpoint with a customer from over 15 minutes to almost instant and significantly reduced the time for the agent to phone. We could prioritise that for the agent because we could read the communication from the customer and identify where they came from.”

AI can digitise and process invoices across different formats by combining natural language processing with computer vision to read the text inside a purchase order PDF or a scanned handwritten document. It can also automate accounts payable and improve forecasting in finance through data-led predictive models, freeing finance experts for value-added work, like interpreting and acting on the outputs.

‘‘I'm in the middle of working with somebody from our data science team to try and work out how we can bring in multiple assumptions to improve our forecasting,” said one attendee. “So, you've got a multi-faceted type of linear regression that's then going to come up with multiple possibilities.”

The first challenge many businesses encounter when attempting to roll out AI is approaching pilots with the wrong mindset. "Pilots should always have the intention of being production-ready. If you do a pilot just to pilot something, you'll often struggle to overcome the hurdle. You buy into a pilot with the intention of going into production," says Harrison.

One attendee also pointed out that organisations should consider their processes before adopting any new technology. "The key thing for me is you've got to step back and look at your process. You've got to look at why you need it and what you need to improve before you adopt the technology," he said.

“Cultural change is a big part of it because that will decide whether the pilot succeeds or fails. Having that plan enables you to work with the business, work with the stakeholders, and say, ‘This is what we're planning to do, this is how we're going to do it, and this is how it's going to make your life better,” he added.

Another key challenge is uniting different units and sites within the business, which may all have unique ways of working. Harrison recommended transforming how the function runs in tandem with technology implementation in instances like this: "There are two ways to do it. We transform and then implement technology, or we use the technology to lead the transformation."

Most roundtable attendees agreed that talent shortages and change management were the primary challenges in adopting and scaling AI within their organisation.

Harrison recommends that business leaders approach it by covering a triangle of 'awareness, understanding, and expertise.' Employees should be aware of what the company is trying to do and how they hope to achieve it. Key figures with an understanding of the technology can be tasked with championing the initiative, and finally, the expertise at the top will deliver the capability.

“If you haven't got that triangle right, and you're top-heavy, you can have all the expertise in the world, but no one's going to adopt it because they're not aware. No one understands what you're trying to do. You can make everyone aware of AI, but if no one's there to deliver it, they'll easily get put off and move on to the next thing. That triangle is important to start your journey to implementing AI more successfully," says Harrison.

As AI becomes more prevalent in the workplace, the existing skills gap may widen. One attendee suggested empowering employees to embrace learning to help retain their competitive edge in the talent market and within the company.

"It's a real opportunity to tell people, 'Come on this journey with us. You'll learn these skills. You may not even use them within our business for years, but you'll be well equipped to go and use those skills in another business, and you're going to be valuable to people in the team who are willing to go on those journeys with you,'" she said.

“Ultimately, it's about upskilling and training. And where that happens is very organisationally dependent,” says Harrison. “If you're a fragmented organisation, then in-function or in-division normally works better. If you're an organisation that's very centralised, then centralised is normally the answer.”

Governance frameworks and a code of ethics should also be critical considerations in any AI implementation journey. Not only that, but they must also be continuously updated to keep up with rapid technological advancement. One attendee said, "Before this meeting, I thought, 'Oh, my company has an AI policy. I'll read it.' It's 15 months out of date — a lot has changed in AI in 15 months."

It's also critical to monitor for biases that might be baked into the AI engine and to be transparent about the data and assumptions used in AI models. Harrison recommends taking a contextual approach in all situations, saying, "You have to have the metrics available to monitor for bias constantly, but you shouldn't use those metrics in situations that don't require them."

Maintaining sight of your company's ethics and values is also advisable when presented with the myriad opportunities that AI can deliver to your organisation. Governing bodies are often too slow to match the pace of technological advancement, so the responsibility falls to the business regarding what is and isn’t acceptable.

“From a regulatory perspective, it's always slow. I think it's about how you govern your own organisation from an ethical perspective that is going to drive a lot of use of AI, particularly in the next couple of years,” Harrison says.

Read more: 5 key ways Gen AI is impacting businesses

For more insights on AI digital transformation and its many business applications, visit our Future of Work hub. To learn more about hiring and skill sets for the coming year, download a free copy of the 2025 Robert Half Salary Guide or contact our team.

Top reasons to resign from a leadership position

IPO readiness: Preparing the executive board and why chemistry and experience both matter

How to make to make flexible working for parents work for you