AI moves from hype to targeted frontline & finance use

Business leaders and technologists are marking AI Appreciation Day with a more sober view of artificial intelligence and its role in frontline work, finance and cyber security. Executives from WorkJam, Aqilla and HackerOne describe a shift from experimentation to targeted deployments that combine automation with human judgement.

The comments reflect a wider reassessment of AI projects. Organisations continue to invest, but face closer scrutiny over costs, returns and the risks of opaque decision-making. Vendors and customers now talk less about broad transformation and more about specific use cases in functions such as retail operations, accounting and security testing.

In frontline-heavy sectors such as retail and hospitality, WorkJam sees AI as a way to streamline daily tasks rather than replace staff. Mark Williams, Managing Director, EMEA, at WorkJam, argued that the most immediate impact will be on workers who rarely feature in boardroom AI strategies.

"While much of the AI conversation has focused on desk workers, some of the greatest opportunities lie on the frontline. Employees in retail and hospitality, for example, often spend valuable time searching for information, waiting for answers from managers and completing repetitive administrative tasks that take time away from serving customers.

Williams pointed to AI agents embedded in frontline operations platforms that respond to staff queries through natural-language interfaces connected to structured operational data.

"AI agents can reduce this friction by acting as intelligent assistants for employees and managers alike. Employees can ask natural-language questions to AI chatbots integrated into frontline operations platforms and connected to operational data. The AI assistant can respond instantly and accurately, providing the necessary training assets or advising whether staffing levels would allow a leave request on specific days, for example. This ensures employees get the answers they need quickly, while reducing the burden on managers to respond to routine queries," said Williams.

He added that workflow automation is beginning to link inventory systems with task management tools for store staff.

"AI can also trigger workflows that automatically create tasks for employees, helping improve productivity and prioritise work as needed. For example, instead of simply notifying an employee that stock is running low, connected inventory systems can trigger an AI workflow that creates a replenishment task within the frontline operations platform. The workflow assigns it to the right employee based on location and workload, includes all relevant product information and delivers it directly to the employee's mobile app," Williams said.

Williams described so-called agentic AI as an emerging area and cited WorkJam research indicating that only one in five retailers view their AI maturity in workforce optimisation as advanced or mature.

"Although most businesses are still in the early stages of their AI journey, agentic AI represents the next frontier. Our recent research found that only one in five retailers describe their AI maturity for workforce optimisation as advanced or mature, highlighting how much opportunity still lies ahead. Unlike better-known generative assistants, agentic AI can work autonomously, making decisions, taking action and adapting if its initial approach is unsuccessful, without constant human intervention. For example, a low-stock alert would not just trigger the creation of a task for an employee to action. These agents could coordinate the replenishment process by checking supplier availability, recommending alternative products if the preferred item is unavailable and escalating only the final purchase sign-off for human approval," Williams said.

He said the value for frontline operators now centres on easing day-to-day decision friction in tight economic conditions.

"For frontline organisations, the greatest value of AI is removing friction from the thousands of operational decisions made every day. As economic pressures grow, AI embedded in frontline operations platforms will increasingly become an invaluable tool for improving frontline execution, helping businesses do more with less while delivering the consistent customer experiences that set them apart," Williams said.

In finance and accounting, Aqilla sees a similar pattern, with automation taking on repetitive work while senior staff retain control over judgement-led decisions. Hugh Scantlebury, Chief Executive Officer and Founder of Aqilla, said the discipline is well suited to structured automation but warned against treating AI as a decision-maker.

"Finance and accounting is often considered the ideal environment for AI because so much of the work follows structured, rules-based processes. To an extent, that's true. AI is helping teams process invoices faster, automate reconciliations and generate month-end reports in a fraction of the time. That saves time and money, while freeing accountants from repetitive work so they can focus on the strategic thinking that really adds value," Scantlebury said.

He stressed that finance remains interpretive at its core and relies on experience and context.

"But efficiency is only part of the story. At its core, finance is about making decisions informed by numbers, and that's where human judgement is, and will remain, essential. AI can support that process, but it can't replace the experience, context and judgement that finance leaders bring. Think about it like this: AI might recommend that a charity struggling with cash flow reduces expenditure to strengthen short-term finances. But trustees also have to consider what that decision means for the people who depend on those services. Those wider considerations are matters of judgement, not simply arithmetic. Equally, a growing business may decide to invest in a new market despite short-term financial pressure because management understands the long-term opportunity. AI can model the risks, but it cannot weigh ambition or strategic intent in the way experienced leaders can," Scantlebury said.

He said organisations that keep people at the centre of financial decisions stand to benefit most.

"The real opportunity is not to replace human judgement with AI, but to combine the strengths of both. Finance has always been about more than producing accurate numbers - it's about understanding the story they are telling and deciding what to do next. Organisations that use AI to remove manual work while maintaining strong governance and keeping people at the centre of decision-making will be the ones that realise its full value," Scantlebury said.

Security leaders describe an even more adversarial landscape as attackers and defenders adopt AI tools. Kara Sprague, Chief Executive Officer of HackerOne, highlighted the asymmetry between automated attack surfaces and the need for human-led validation.

"AI Appreciation Day tends to celebrate what AI creates. In security, the more interesting story is what it defends against," Sprague said.

She said threat actors were early adopters of AI and now use it to speed up the exploitation of weaknesses.

"Attackers adopted AI first. They use it to find and exploit weaknesses faster than any human team could match, and that is exactly why AI has become indispensable to defenders. But AI has not replaced the human security researcher. It has raised the value of one. In the first half of this year, security researchers earned roughly $47 million through our platform, up 25% from the same period a year ago, even as AI made finding vulnerabilities faster than ever. AI can flag a possible vulnerability in seconds. It still takes human expertise to prove that vulnerability is real, understand how an attacker would actually use it and decide what to fix first. Machines generate the possibilities. People confirm the truth," Sprague said.

She said the most effective security strategies now combine automated scanning with expert review.

"The organisations getting this right are not choosing between security researchers and AI. They are pairing them. Speed from the machine, judgement from the human. That is what continuous security now demands," Sprague said.