Friday, 4 September 2026
Artificial intelligence could automate or support the majority of routine finance processes within the next several years, changing how CFO teams handle forecasting, financial close, tax, treasury, compliance and reporting, according to a presentation at AI4 2026 examining SAP’s vision for the finance organisation of the future.
Speaking during the session “Finance Reimagined: How AI is Automating Every Finance Function With SAP,” the SAP presentation argued that finance departments are moving beyond isolated machine-learning applications and generative AI tools towards a model in which specialised AI agents work across interconnected financial processes.
SAP calls this direction “autonomous finance”. The concept does not envisage finance departments operating without people. Instead, AI assistants and agents would perform more of the repetitive analysis and processing while employees remain responsible for oversight, judgement, approvals and higher-level decisions.
The shift comes as finance departments face growing external pressures. Regulatory requirements are becoming more complicated, including the continuing expansion of electronic invoicing regimes, while companies also have to react more quickly to tariffs, geopolitical disruption, supply shortages and other unexpected economic developments.
Talent represents another challenge. Experienced finance professionals will continue to leave the workforce through retirement, while younger employees are entering organisations with different expectations about the type of work they want to perform. Routine processing, repetitive reconciliations and lengthy manual reporting may become increasingly difficult to justify when technology is capable of carrying out a growing share of those activities.
At the same time, removing repetitive work creates its own problem. Junior accountants traditionally gained experience through precisely the tasks that AI may increasingly perform. If young employees no longer spend years manually processing transactions and preparing financial information, businesses will need alternative methods of developing the judgement required by future controllers, finance directors and CFOs.
The presentation suggested that mentorship could therefore become more important rather than less. Young professionals may need to work more closely with experienced employees so that judgement, context and business understanding can be transferred deliberately instead of being accumulated gradually through repetitive work.
This becomes especially important as companies introduce agentic AI into processes where a human remains responsible for the final decision. The more work delegated to machines, the greater the importance of ensuring that employees reviewing their recommendations understand what a correct result should look like.
SAP’s broader proposition is that a large proportion of finance processes can eventually be fully or partly automated with AI. The presentation put the potential figure at around 80%, although this should be understood as SAP’s view of the opportunity rather than an independently established measure of current automation across businesses.
The commercial consequence could be substantial. Finance teams traditionally spend considerable time collecting data, preparing reports, reconciling accounts and explaining historical results. If AI performs more of that work, employees can devote greater attention to planning, financial analysis, scenario testing and advising senior management.
This could accelerate the transition from periodic financial management towards continuous financial management. Planning and budgeting are one example. Annual budgets and quarterly forecasts exist partly because producing and updating them requires substantial effort. If AI agents can continuously process operational and financial information, companies could increasingly move towards rolling forecasts that adjust as business conditions change.
That would make finance less dependent on reporting cycles and potentially allow companies to respond more quickly to changes in sales, costs, working capital, currencies or external market conditions.
The presentation also demonstrated how this could change the financial close. Instead of employees manually reviewing every clearing item or reconciliation, specialised agents can analyse transactions and prepare proposed actions for human approval. An accounts receivable agent, for example, could identify transactions that appear suitable for clearing, while an intercompany reconciliation agent could compare receivables and payables between entities and highlight exceptions requiring attention.
Journal entries could similarly be prepared for review rather than constructed completely manually. The important distinction is that the accountant remains involved. In the demonstration, employees reviewed the agent’s reasoning and approved proposed actions before they were posted.
Once routine closing activities had been completed, the same AI environment could move into financial analysis, assessing measures such as gross margin, liquidity, returns and expense ratios.
This combination of execution and analysis illustrates a larger change occurring in enterprise software. Rather than employees moving between multiple applications and extracting information themselves, conversational interfaces are increasingly becoming a control layer through which users instruct software to retrieve information or perform tasks.
SAP’s Joule platform is designed around this concept. Instead of functioning simply as a chatbot, it is intended to connect users with financial information, applications and specialised agents through natural-language instructions.
For finance departments, the difference is important. A general-purpose chatbot may answer questions or generate text, whereas an enterprise agent connected to financial systems could potentially analyse transactions, apply business rules and initiate workflows.
That capability also raises governance questions. Financial systems require considerably stronger controls than many everyday AI applications because errors can affect company accounts, tax positions, payments and regulatory reporting.
SAP is therefore placing significant emphasis on auditability. The presentation said activity performed by AI agents can be recorded alongside information identifying the agent, the assistant involved and the employee who reviewed or approved the action.
Taxation is another area where greater automation could have an important financial effect. Tax departments frequently depend on information produced elsewhere in an organisation. Delays in receiving that information can reduce the time available to assess liabilities, structure transactions or plan efficiently.
Giving tax teams faster access to relevant information could therefore provide benefits beyond reducing administrative work. The larger advantage could come from allowing tax specialists to make decisions earlier.
Treasury presents similar opportunities. AI systems could monitor liquidity, cash transfers, currency exposures and working capital against predefined policies, escalating exceptions to employees rather than requiring people to examine every transaction manually.
Compliance monitoring could also become more continuous. The presentation cited an example involving Pfizer, where AI was described as supporting internal-control monitoring across the organisation rather than relying exclusively on periodic reviews. Such examples illustrate how finance controls could gradually move from retrospective sampling towards more continuous monitoring.
Reporting itself may also change. Companies often continue producing reports because they have historically been requested rather than because management still uses them. AI-powered self-service tools could allow authorised employees to retrieve or generate information when required, reducing the need for finance departments to maintain large catalogues of recurring reports.
One organisation referenced during the presentation was said to have removed approximately three quarters of its previous reports after introducing greater self-service capabilities. The figure should be regarded as an individual case study rather than evidence of a typical outcome across companies.
The implications for CFOs extend beyond technology expenditure. Boards are increasingly likely to ask finance leaders what return companies are receiving from their investment in AI. Some benefits will be measurable through lower processing costs, shorter close cycles or reduced manual effort.
Other returns will be less straightforward. Giving skilled finance professionals more meaningful work could improve employee satisfaction and retention. Faster information could improve management decisions without producing an easily identifiable cost saving. Better control monitoring could reduce risk even if it does not immediately increase reported profit.
The challenge for CFOs will therefore be developing measures that capture both direct financial returns and improvements in organisational capability.
SAP’s position is also that AI transformation should not automatically become a multi-year corporate programme. The presentation recommended beginning with a specific problem. Finance leaders could ask experienced employees which activities consume significant amounts of time while providing relatively little professional value.
Companies could then select one suitable process, introduce automation, evaluate the result and progressively move into other areas. This reflects a broader message emerging from AI4 discussions on finance: businesses should begin with an operational outcome rather than deciding they need AI and then searching for somewhere to use it.
The longer-term change, however, could be much larger than individual use cases. If financial close, forecasting, treasury, tax, compliance and reporting increasingly become supported by connected AI agents, the traditional finance organisation may gradually shift from producing information towards supervising a continuously operating financial system.
That would not necessarily eliminate the CFO organisation. It would change what people inside it are expected to do.
The finance professionals who remain most valuable would increasingly be those capable of interpreting results, challenging AI recommendations, understanding controls, communicating with operating teams and translating continuously updated financial information into business decisions.
The ultimate destination is therefore not finance without people. It is finance in which considerably less human effort is spent assembling information and considerably more is spent deciding what the organisation should do with it.
Source: CIJ.World Research & Analysis Team
