Why Computacenter is resisting rushing into agentic AI

Automation lead argues that enterprises need to identify worthwhile use cases before embracing the next phase of AI

Despite growing enthusiasm around agentic AI, Computacenter says it is deliberately taking a cautious approach.

Drawing on lessons from more than a decade of automation projects, the reseller says it is taking a measured approach to agentic AI adoption, focusing on governance, trust and business value rather than chasing the latest technology trend.

CRN sat down with Kasim Iqbal, automation lead at Computacenter, to discuss the organisation’s AI journey and why rushing into agentic AI is not something it plans to do.

What has Computacenter’s agentic AI journey been like so far?

“We’re still starting off.

“We are using AI within Computacenter. Before we go out to our customers, we’re using it ourselves. We’re using AI for voice, unstructured data, Copilot is another thing that we’re heavily using within our organisation.”

Computacenter is using AI internally to develop expertise, best practices and governance frameworks that it can apply in customer engagements.

“We can showcase what we’ve done for ourselves, our learnings, the best practices we followed and the challenges we have faced.

“If a customer comes to us with a scenario that we haven't tackled, we'll take that away, and see how best to tackle that.”

Although, when it comes to agentic AI and automating processes with the technology, Computacenter is taking its time.

“But when it comes to RPA and AI, we’re not doing that.”

He explained that it is all about finding the right tool for the right application, and with the majority of use cases, traditional automation processes are continuing to work effectively.

Although, agentic AI is on the roadmap.

“We don't want to just be targeting AI because it's a buzzword.

“We want to make sure that the use case that we target has actual business value, and that's just purely because we've been doing RPA for the last ten years.

“When RPA started off, it was similar to AI. Buzz was everywhere, everyone wanted RPA, and sometimes we didn't target the right use cases.

“We don't want to repeat those mistakes.”

Computacenter is taking those learnings and applying them to AI, taking time to make sure it is finding real value before implementing the technology.

“We don't want to just follow the market because it's a lot of investment, time and effort.

“We want to make sure we spend that on something that will give us a future prospect, future ways of working … we want to make sure there's actual value.”

Where are the valuable agentic AI use cases for the channel?

While AI continues to dominate boardroom discussions, many organisations are still struggling to identify high-value use cases.

As Iqbal has said, RPA continues to show valuable returns. However, monetising AI is still in its early days.

“There’s a lot of space for AI within unstructured data, such as email manipulation and data manipulation.

“It can help make decisions quicker than a normal human would make.

“That's an aspect that AI can really be powerful in, and over the years, AI is getting smarter, so it will really make a difference and will make decisions faster, and can really help operations deliver outcomes quicker.”

To find the biggest gains, organisations must first look at the low-hanging fruit.

“If an organisation is starting off, [looking at] any repetitive processes is standard.

“Then if you've tackled all of them, start going out to the business operation reviewing processes.

“It's all about looking at the repeatable processes, and that's where you start off with agentic AI.”

Will AI change the role of traditional automation?

“My personal view is I don’t think AI will replace RPA,” Iqbal said.

He explained that organisations that can merge the two is where the best outcomes and value will lie.

“For example, using AI for the decision-making, and then passing that onto RPA.”

AI is still far away from meaningfully replacing humans and RPA for truly valuable use cases.

While it has its place, the technology needs to be able to be controlled, and customers need to be able to trust it within their workflows.

Being able to bring customers on the journey is one of the big hurdles of making it work.

“You have to be able to show the outcome and the real value, which automation can do, but AI is still in its early days.”

Misconceptions about automation

When it comes to automation, whether traditional or powered by AI, the industry still has misconceptions.

“The biggest one is that RPA is dead,” Iqbal explained. “I think there is still a place for RPA.”

“We’re an IT company, and RPA within the IT industry is not the best serve.

“But even when we started the journey, we didn't know where we would will end up, but after we did the first use case, we started looking at operations, and we saw so many repetitive activities within the operations.

“In the last ten years, it's given so many hours back to the operations, and there's so many more operation processes that are repetitive and rule-based.

“So this term ‘RPA is dead’ or ‘AI will kill RPA’, I think it's not going to happen, not anytime soon.”

AI has the potential to significantly reshape how organisations operate.

But before that happens, organisations need to look internally and make sure they are ready for the transformation.

Iqbal explained, “AI is powerful, but it's only as powerful as the data it has.

“If you give it bad data, the decision that it will make will be, too.

“Whereas, automation, as long as the process is structured, you will be able to automate it.”

Education, governance and planning for AI

A lack of education and meaningful planning is causing organisations to go astray.

Organisations are aware they should be implementing AI, but aren’t sure where in their businesses they should be doing so to find real value, leading to fragmented implementation and failed pilots.

“Many organisations don't know which direction to take the AI journey,” he explained.

“I think any organisation that just follows AI because it's a buzzword in the market aren’t going to get very far because there's no viable use cases.

“You [also] need a strategy on where you are now and where you want to take that journey.”

The journey doesn’t start and end with education and strategy, as governance is a key hurdle impacting adoption.

“It all comes down to governance.

“Governance is slowing down the use of AI.

“Governance and trust is something that we want to make sure we have governance and control over the stuff we are doing with AI within our organization.

“We have an AI governance board and strategy board that decides what AI, and what tooling we use.

“But even then, it's all about having control, making sure that the outcome that AI is giving is trustworthy, is accurate, and it's not just giving the wrong message.”

For Computacenter, the immediate focus is not deploying agentic AI at scale, but building the governance, trust and use-case maturity needed to support it when the technology is ready.