The BOSS, Automation and AI

In summary

• The Business Owners Sentiment Survey shows many firms cutting costs through automation, but there are other powerful drivers
• In adopting automation and AI, however, businesses need to be led by their needs, not the technology 
• They also need to professionalise their approach, and not let their ambition outrun their expertise

S&W’s recent Business Owners Sentiment Survey showed many firms looking to cut staffing costs. Stung by rising wages and, particularly, the increases in employers’ NI in last Autumn’s Budget, we found companies looking at redundancies, hiring and pay freezes, and moving jobs abroad.  

Automation, too, was firmly in business owners’ sights.  Half of employers told us planned to increase the use of automation to replace people as a result of the hike in NI contributions.  

While cutting staffing costs may be a common motivation for automating processes, it’s far from the only one, however. That range of motivations means an appetite for automation isn’t going away, whatever happens to employment costs.  

According to Director Adrian Hextall in the digital services practice at S&W, discussions of automation often overlap with those about artificial intelligence (AI) and the distinctions between the two are not always clear in business owners’ minds. 

“A lot of the work tends to be automation, replicating existing processes, rather than creating journeys where the solution thinks for itself, which would be AI,” says Hextall. And in most cases he’s worked on recently, the primary driver has not been to cut costs but to ensure accuracy and insight.  

For finance functions, for example, he’s recently worked with clients on VAT reporting – replacing Excel workbooks for VAT returns with an automated process that brings significant benefits in terms of accuracy. 

“We’ve found old Excel workbooks peppered with logic consolidation and cross-casting errors that go unchecked until HMRC reviews them,” he says. “Rebuilding these automation routines significantly increases the accuracy. They can analyse the entirety of transactional information, checking it every quarter, while current manual processes favour sampling that can easily miss errors since only a small proportion is ever checked.” 

It has reduced work, but it’s work that was already outsourced. 

“It’s a cost saver in the sense that you’re no longer paying a third party to do the work manually. You can bring it all in-house with an automation routine and manage it with a much smaller finance team than you’d otherwise need,” Hextall explains.  

Interest in AI is also driving discussions, and our survey found significant concern about competition from businesses harnessing the new technology. Among a list of issues threatening businesses, it ranked just outside the top three.  

As a result, there’s significant pressure in some sectors to engage with the new technology, says Partner Laurence Kiddle. 

“There’s an element of keeping up with the Joneses,” he says. “Companies need to be seen to be investing in AI, because otherwise their investors don't take them wholly seriously. 

“No one wants to get left behind.” 

One result of this, however, is a proliferation of failed AI programs and pilots. MIT research published in the Summer showed that 95% of generative AI pilots by companies did not deliver a return on investment.  

AI tools are designed to be user friendly and intuitive for the layman, but what works for planning a holiday itinerary is not necessarily suitable for solving business issues.  Despite this, many organisations encourage a “cult of the amateur”, with projects driven by non-specialists using free tools. That can not only limit the impact but also increase risks. Many businesses are failing to control the information being shared with AI tools such as ChatGPT. In some cases, that will include sharing GDPR-protected data with public Cloud services.  

“I suspect there’s a lot of unconscious data breaches with AI and large language models,” says Kiddle.  

That’s not to say there’s not significant potential for AI. There is, but the conversation should start with the business problem rather than the technology, and solutions should employ the processes and skills expected when developing other business-ready tools.  

“Everyone’s keen to talk about new AI tools, but that’s approaching the issue from the wrong end,” says Kiddle. Instead, businesses should focus on what they want to achieve and then look to the technology to solve these issues, whether that’s AI or more traditional process automation.  

“It’s really going back to basics and looking at what you’re trying to do as an organisation and what’s holding you back from that.” 

That probably means looking at three areas, according to Hextall:

• Time sinks – Tasks and processes that take significant amounts of time on processes that could be done by bots, data processing routines or AI 
• Blind spots – “Areas where there’s some insight that’s missing from your commercial understanding or management information, which you could generate by looking at the data a different way,” as Hextall explains 
• Errors and mistakes – Where existing processes cause repeated problems through inaccuracy or other issues that could be corrected through automation or AI 

In practice, this means that some of the more successful current uses of AI may seem unspectacular: automating and enhancing back-office processes for improved efficiency, accuracy and insights.  

“It’s not glamorous, but we’re still in the early days of AI,” explains Kiddle. Just as the early days of computers in the finance function saw them start with running payroll, many of the first applications of AI will be on similar “dull but necessary” tasks. 

“You’ve got to start somewhere.” 

There are two key advantages to this approach, too. First, it enables an organisation to hone its approach to automation and AI – and managing the risks – in the back office, away from more commercial and client-facing value drivers, where mistakes can significantly damage reputations.   

Second, focusing on such areas makes for a more attractive working environment. These are not tasks that make employees excited to get out of bed in the morning, Kiddle points out.  

“Automating some of these dull processes can help you build a happier, more engaged workforce – whatever size it is.”