🏠︎ | Past Sessions | Bridging the Gap How Agentic AI can Empower Employees and Transform Finance Teams
- Event: Finance Forum 25
- Date: 7 October 2025
- Speakers: Pedro Batista, David Atherton, Carolina Einarsson
- Estimated read time: 7-8 minutes
Quick summary
This session explored how agentic AI is beginning to change the way finance teams operate, without removing control from CFOs.
The discussion focused on why adoption has lagged behind intent, how finance leaders are using AI as an assistant rather than a replacement, and where early value is already being created across finance operations.
A recurring theme was that progress depends less on advanced technology and more on mindset, trust, and practical experimentation. Teams that start small, keep humans in the loop, and focus on real workflow pain points are seeing time savings and stronger collaboration across the business.
The gap between AI ambition and reality in finance
The session opened with a familiar tension for senior finance leaders. AI is widely recognised as important, yet adoption remains cautious.
David Atherton framed the gap as a combination of fear and complexity. Concerns around compliance, data readiness, security, and trust make it difficult for finance teams to see a clear and safe path forward. AI is not short of potential use cases, but understanding where it can be adopted quickly and responsibly is less obvious.
Carolina Einarsson added that this caution is rooted in the CFO’s responsibility for control. Finance leaders are accountable for accuracy and governance, so the idea of handing processes to AI can feel uncomfortable. Her framing was pragmatic, AI should be treated as a finance assistant, not an autonomous decision maker. The value comes from augmentation, with a human layer still validating outputs.
“You should treat AI as a data analyst or a finance assistant, not something that replaces judgement.” Carolina Einarsson, CFO, Essentia Analytics
Agentic AI as an enabler, not an extra burden
A key risk highlighted in the discussion was overwhelming teams with change. Atherton stressed that adoption fails when AI adds complexity rather than removing it.
The most effective implementations focus on automation and approvals that sit naturally within existing workflows, such as expense management, purchasing, and day to day operational finance tasks. When AI supports what teams already do, rather than asking them to learn entirely new skills, adoption is far more likely to stick.
Einarsson shared a small team perspective, where finance is frequently asked for operational answers, from revenue performance to customer renewals. She described the potential for agentic AI to surface trusted answers directly to employees through tools like Slack, reducing interruptions while making finance more accessible across the business.
Early wins are practical, not futuristic
Despite the caution, the session made it clear that agentic AI in finance is already delivering tangible benefits.
Atherton pointed to reconciliation as one of the most valuable early use cases. Automated matching across procurement, invoicing, and payments removes a major manual burden and allows finance teams to focus on analysis rather than transaction processing.
Einarsson described how AI has accelerated her use of existing tools rather than replacing them. By combining AI with spreadsheets, scripts, and automation platforms, she has built simple systems that would previously have required specialist development skills. The time saved is then reinvested in reviewing outputs and understanding what the numbers are really saying about the business.
The common thread was that AI works best when paired with a clear source of truth and a defined problem. Automation reduces manual effort, but judgement remains firmly with finance.
Build versus buy, and why it is rarely either or
The panel explored whether finance teams should build their own AI solutions or rely on vendors.
Einarsson’s view was that, for most finance teams today, buying and extending existing tools is the most efficient path. Different AI tools excel at different tasks, from formula generation to visualisation, and experimenting across them helps teams understand where value really sits.
Atherton echoed this, noting that many platforms are embedding AI directly into their products. Rather than reinventing the wheel, finance leaders should focus on adopting these capabilities and ensuring they integrate cleanly into their data architecture.
Both speakers agreed that fully bespoke AI may become relevant later, but for now progress comes from selective, practical adoption.
Collaboration improves when finance becomes easier to work with
One of the more human insights from the session was how agentic AI can change perceptions of finance.
Einarsson described finance as historically process driven and sometimes intimidating. By introducing AI agents that answer questions or assist with models, finance becomes more approachable and collaborative. She shared an example of building an AI agent alongside her CTO during a company hackathon, highlighting how no code tools allow finance leaders to participate directly in innovation.
Atherton added that AI also helps provide context, not just numbers. By linking planning, execution, and retrospective analysis, finance can better explain why outcomes occurred and how decisions performed over time.
What good looks like, practical actions for finance leaders
This session translated into several concrete actions for CFOs and finance transformation leaders considering agentic AI.
Questions to ask your team- Which manual tasks consume time every week and add little insight
- Where delays are caused by chasing data rather than analysing it
- Which decisions would improve if answers were easier to access
- Where human judgement must remain explicit and documented
- Finance time shifts from processing to analysis
- Teams trust AI outputs but still validate them
- Other departments engage finance earlier, not just at sign off
- Tools feel embedded in workflows, not bolted on
- Treating AI as a replacement for accountability
- Adding tools without clear ownership or data governance
- Expecting immediate transformation rather than incremental gains
- Ignoring staff confidence and change readiness
What good looks like in practice
Agentic AI empowers employees when it quietly removes friction. Finance retains control, but gains speed, capacity, and better conversations with the business. The result is a finance function that supports decisions rather than simply reporting outcomes.
Conclusion, curiosity beats certainty
The session closed with a simple but powerful message. Progress with agentic AI does not require a perfect roadmap.
Atherton emphasised the value of peer learning and shared experimentation. Einarsson encouraged finance leaders to stay curious and use AI regularly, even in small ways. The teams that move forward are not those with the most advanced tools, but those willing to explore, test, and learn while keeping judgement firmly in human hands.
Speakers
CEO UK and VP Payments and Operations, Payhawk. Leads UK operations and oversees payments and operational strategy at Payhawk.
Country Director UK, Staria. Works with finance teams across multiple markets to scale finance operations and systems.
CFO, Essentia Analytics. Finance leader with experience building scalable finance functions in growing technology businesses.