How AI agents help drive a new finance operating model: What CFOs need to know

• Up to 90% time savings in key processes
• Up to 60% of team time redirected to insight work
• Up to 40% improvement in forecasting accuracy and speed

You’ve been working to get here for 15+ years

If your finance function is like most, over the last decade a key benchmark, finance cost as a percentage of revenue, has driven your strategy:

• Using technology to reduce transaction and processing costs
• Deploying global and onshore business services models using that technology
• Connecting that technology to user-driven tools at the edge to generate insights

According to PwC’s 2024 Finance Effectiveness Benchmarking Report, leading finance teams have achieved that goal. They're spending more time generating insights and less time on automatable tasks, while reducing their costs (as a percentage of revenue) by nearly 25%.

Finance has become leaner and more insight-driven, but expectations to do more with less are rising fast. In many organizations, traditional levers, such as cloud-based ERPs, integrated data platforms, automation and edge analytics, have hit their limits. You’re digital. You’re efficient. Yet you can't keep up with the increasing demand from the business, regulators and investors.

The next breakthrough for finance isn’t another system or tool — it’s a new way of working. Across the business, AI agents are helping reshape how work gets done, and finance is at the forefront of this evolution. By pairing human expertise with AI agents, finance can scale capacity, accelerate insights and stay ahead of rising demands. With a strong digital foundation already in place, the opportunity is clear. What’s required now is the leadership to move fast and capture it.

^{Source: PwC’s 2024 Finance Effectiveness Benchmarking Report}

How AI agents are transforming finance

AI agents make a new finance operating model possible because they can act intelligently, autonomously and in teams. As you would with your human workforce, you typically give each AI agent a different role. One might be an accountant, another an FP&A analyst, and a third a compliance specialist. Every agent has the skills and data sets to match its role.

Next, you orchestrate the different agents into a workflow and give them instructions, like “reconcile invoices with purchase orders (POs)” or “consolidate cash positions and forecast inflows/outflows.” They’ll work together to get the job done. Every AI agent can recall what it did before and what the outcomes were. It learns from your inputs, reviews and exception-handling how to better manage the next set of data and tasks. Like a human gaining experience, an AI agent can keep getting better at its job and create new solutions.

Given our expertise with implementing AI agents and transforming work, we’ve conducted a detailed analysis of the finance function, enabled by PwC's workforce AI task analyzer. We looked across more than 40 processes spanning procure-to-pay, order-to-cash, record-to-report, financial planning and analysis, and treasury. Our analysis shows that finance tasks tend to fall into three categories: human-led, agent-assisted or fully agent-driven.

The takeaway? AI agents can independently operate — with the right deployment and governance model — nearly every aspect of shared service centers operations. In centers of excellence (CoE), they can assist people with nearly all of their work. For corporate and business finance teams, AI agents can augment strategic guidance, customer-facing finance functions and more. Whatever the task, AI agents can free your people from structured, repetitive work, providing capacity to create insights that help fuel higher-value contributions.

^{Source: PwC analysis}

Reimagine finance: PO transaction processing and matching

This is what an AI-driven workflow looks like: In PO transaction processing and matching, agents do nearly all the work — fast, accurately and at scale. They can reduce cycle times by up to 80% while improving audit trails, reducing compliance risk and enabling scale without added cost. Here, agents can deliver value by automating repetitive tasks so people can focus on analysis, strategy and vendor management.

When a finance professional submits an invoice for review, one AI agent extracts key information from the invoice. A second pulls the relevant contract or master services agreement (MSA). A third compares the invoice to the contract terms and flags any discrepancies. A fourth agent drafts an email to request resolution, credit or clarification. Only then does a finance specialist step in, to approve or edit the draft email or escalate the issue if needed.

While the AI agents handle this repetitive work, your people can focus on higher-value tasks. They’ve been collaborating with other AI agents and colleagues to review vendor performance, investigate recurring overcharges and non-compliant invoices, renegotiate contracts or MSAs, and improve contract intake processes. The result isn’t just greater efficiency. It’s real cost savings and stronger vendor performance across the board.

Procure to pay: How AI agents can help automate the process and free up finance specialists

AI agents streamline and enhance procure-to-pay workflows — automating transactional steps like invoice validation and discrepancy checks — so finance professionals can focus on higher-value activities such as contract optimization, vendor performance and policy enforcement.

Reimagine finance: Treasury cash positioning and forecasting

In more complex finance tasks too, agents don’t replace your people. They work alongside them. Once again, they handle the routine steps, so your team can focus on the decisions that matter.

Take treasury operations. Agents can pull and consolidate cash balances, predict near-term inflows and outflows, flag potential surpluses or shortfalls, recommend transfers or investments, log completed actions and refine forecast models.

With forecasts in hand, your team can sharpen capital allocation strategies, adjust cash thresholds and update investment policies based on patterns the agents uncover. Using agent-generated cash position analyses, they refine strategies for pooling, sweeping and internal lending. They also have more time and better data to advise sales and account management teams with greater impact.

Treasury: How AI agents take on the heavy lifting in forecasting and liquidity

With AI agents handling cash pulls, forecasts and recommendations, treasury teams can focus on what’s next.

How to build an AI-powered finance function today

Invoice processing and treasury operations are just two examples of how agents can automate high-value, repetitive finance work and assist with more complex tasks. They're also starting to reshape areas like PO transaction processing and matching, collection management, journal entry preparation, supplier risk monitoring, liquidity optimization, financial closes and more.

One reason AI agents deliver value fast? They scale easily and finance team members can create and change them. Each new agentic workflow may be able to reuse code and architecture from those that came before. With a mature tech stack, you can see impact in weeks, and stand up an all-new AI-powered operating model within months.

1. Assess your platform, process and delivery model. If you have a broadly connected and adopted data platform, as many finance functions do, plus a shared services or outsourced delivery model, you can create value with AI agents within 30 days. If not, ROI can still come rapidly, but you may need a little more prep first.
2. Build modular and reusable. Since AI agents can do so much, it's important to seek more than incremental value in silos. If you adopt a modular architecture, it will be easier to reuse code, agents and agent frameworks across workflows — helping you scale quickly and keep costs down.
3. Make orchestration and oversight intuitive. One AI agent can usually do just one simple task. Real value comes from orchestrating agents into workflows with an easy-to-use platform like PwC’s agent OS, which has oversight tools built in.
4. Show your people what's in it for them. Our 2025 Global AI Jobs Barometer shows that AI makes workers more valuable — if they have new skills, like how to oversee AI, and a culture of working in teams with AI agents.