AI in finance is everywhere, but that does not mean every tool carrying an AI label will make a meaningful difference to a finance team.
For CFOs, the important question is not simply where AI can be added. It is what problem the technology is expected to solve and whether the result can be measured.
That is the central point in this Finance Leaders Unplugged conversation between Jon Clarke and Gavin Roe. Both are now Sage Intacct consultants at Inixion, but they have spent much of their careers on the other side of the table. Jon is a former Finance Director and Gavin a former Financial Controller, so their discussion is grounded in the realities of running a finance function.
You can watch the full video here: https://youtu.be/BYcVPrS5Tx8
AI in finance should start with the problem
As Jon says in the video, AI is being talked about everywhere. It is also being added to a growing number of products, sometimes without a clear explanation of what it actually achieves.
That makes it easy for the conversation to start in the wrong place. A finance leader sees an AI feature and then looks for somewhere to use it. A better starting point is the existing problem.
Is the finance team spending too long entering supplier invoices? Are people chasing approvals? Does month-end depend on repetitive checks? Is it difficult to investigate a variance or find reliable information quickly?
Once the problem is clear, the CFO can judge whether AI is genuinely useful. The questions become more practical:
- What task will it improve?
- How much time could it save?
- Will it reduce errors or repeated work?
- Can the team understand and verify the output?
- What will success look like after implementation?
This is the difference between adopting AI because it is fashionable and investing in technology that produces a measurable result.
What practical AI in finance looks like
The most useful applications of AI in finance are often not the most dramatic. They address high-volume, repetitive activities that take people away from analysis and decision support.
Accounts payable automation
Accounts payable is a strong example because the problem is easy to recognise and measure. Finance teams receive invoices in different formats, enter the details, match them to suppliers and purchase orders, check for duplicates and send them for approval.
Sage Intacct AP Automation uses AI to capture bills, create draft transactions, match invoices to purchase orders and route them for approval. The finance team still reviews and approves the work, but less time is spent on manual entry and administration.
The value does not come from the AI label. It comes from reducing processing time, avoiding preventable errors and keeping invoices moving before they become a month-end backlog.
AI-supported month-end close
Month-end is another process where small delays and repeated tasks quickly add up. Finance teams may be waiting for information, tracking progress in spreadsheets and checking the same figures several times.
Sage Intacct Close Automation provides shared checklists, task tracking, notifications and visibility of the close. Its AI-supported capabilities can assist with reconciliations and variance analysis and surface possible issues earlier.
The Close agent helps teams track tasks and identify bottlenecks, but it does not remove human control. Sage states that it does not post entries or make changes without approval. This is a useful example of AI supporting a controlled finance process rather than making financial decisions independently.
Finding unusual activity sooner
Checking data for errors, duplicates or unexpected transactions can absorb a considerable amount of finance time. AI can help by reviewing a much larger volume of activity and bringing unusual items to the team’s attention.
Sage Intacct’s Assurance agent is designed to monitor financial data and flag anomalies, potential errors and reconciliation mismatches. A finance professional can then investigate the item and decide what action, if any, is required.
Again, the practical value is in directing human attention to the transactions that need it most. It does not mean handing accountability to the software.
Getting answers from financial data
AI can also make financial information easier to explore. Sage Intacct’s Finance Intelligence agent is designed to help users ask questions about reports and transactions, investigate variances and obtain explanations from their financial data.
This could make it quicker to answer questions such as why a cost has increased or which customers have invoices beyond a particular age. The answer still needs to be considered in context, especially before it informs an important business decision.
You can read more about these capabilities in Inixion’s guide to Sage Intacct AI capabilities.
Reliable data matters more than an impressive answer
Jon and Gavin are clear that AI output should not be trusted blindly.
An answer can sound confident and still be incomplete, based on poor information or simply wrong. In finance, the consequences of acting on unreliable information can be far more serious than producing an awkward paragraph or an inaccurate meeting summary.
The quality of AI in finance therefore depends heavily on the quality, structure and accessibility of the underlying data. If financial information is fragmented across spreadsheets and disconnected systems, adding AI will not automatically make it reliable.
Finance leaders should ask:
- Which data is the AI using?
- Is that information current and complete?
- Can users trace an answer back to the underlying transaction or report?
- Does the tool respect existing roles and permissions?
- Is there an audit trail?
- What happens when the AI is uncertain?
This is one reason embedded, finance-specific AI can be more useful than a separate general-purpose tool. It operates within the financial management environment, where permissions, transactions and auditability can form part of the design.
Human accountability still belongs with finance
AI can highlight an anomaly, summarise information or recommend where someone should look. It cannot take professional accountability away from the CFO or finance team.
As Gavin explains, there must still be human intervention in key decisions. Users need to understand how a result was produced and be able to go back through the information behind it. Otherwise, trust quickly disappears and the team is unlikely to use the tool.
Human oversight should not be treated as a temporary step that will disappear once people become more comfortable with AI. It is part of responsible financial control.
The strongest model is usually a combination:
- AI handles repetitive processing and reviews large volumes of information.
- The system surfaces exceptions, patterns and possible explanations.
- Finance professionals apply context, challenge the output and make the decision.
That balance allows teams to gain efficiency without weakening accountability.
Invest in people as well as AI finance tools
One of the most important points in Jon and Gavin’s conversation is also one of the easiest to overlook: businesses need to invest in their people as much as the technology.
Buying an AI-enabled solution does not mean a finance team will automatically know how to use it well. People need to understand what the tools can do, where the limitations are and how their own roles and processes may change.
That requires more than a short demonstration at launch. Teams need relevant training, clear guidance and the confidence to question an output that does not look right.
Finance leaders should also involve the people doing the work when considering AI. They are often best placed to identify the repetitive processes, awkward workarounds and control points where automation could make a genuine difference.
How should CFOs assess AI in finance?
The market will continue to produce bold claims about AI. CFOs need a straightforward way to separate useful capability from noise.
Before investing, ask:
What finance problem are we solving?
Define the process, delay or risk first. Avoid beginning with a technology looking for a use case.
How will we measure the result?
Choose a meaningful measure such as invoice processing time, manual transactions, close duration, error rates or time spent investigating variances.
Can we trust and verify the output?
Understand the source data, permissions, audit trail and how users can investigate the result.
Where does human approval remain?
Be clear about which activities AI can support and which decisions still require review and accountability.
Are we preparing the team properly?
Plan for training, adoption and process change. The technology will only create value if people understand and use it.
How Sage Intacct approaches AI-powered finance
Sage Intacct combines financial management, automation and finance-specific AI within the same environment. Its current AI capabilities include support for accounts payable, the financial close, assurance checks and exploring financial information.
This matters because AI is not sitting separately from the finance process. It can work with the transactions, workflows, permissions and reporting already held within Sage Intacct.
The appropriate configuration will depend on the organisation and not every capability will be required by every finance team. A good implementation should begin with the business problem, map the process and agree the outcomes, precisely the approach Jon and Gavin recommend in the video.
For a wider overview, visit Inixion’s AI-driven Sage Intacct hub.
AI should give finance more time to think
The best use of AI in finance is not technology for its own sake. It is removing work that does not need professional judgement so finance teams can spend more time on the work that does.
That might mean less invoice entry, fewer repeated checks, a better-controlled close or quicker access to useful information. The result should be visible in the team’s productivity, accuracy or ability to support decisions.
If the benefit cannot be explained or measured, the AI label alone is not a reason to invest.
If you would like to explore where Sage Intacct AI could make a practical difference in your finance team, request a call back or get in touch with Inixion to continue the conversation.
