Artificial intelligence has rapidly become a strategic priority for finance leaders across industries, fueled by promises of faster processes, sharper insights, and greater operational efficiency. According to a recent KPMG report, 75% of organizations now actively use AI in the finance function, a sharp increase from just 30% in 2024.

Yet for many, the first generation of AI tools has fallen short of expectations. Rather than fundamentally transforming how finance teams operate, AI has largely been confined to chatbot functionality — summarizing information and offering generic suggestions without understanding broader business context or connecting to the systems where finance work actually takes place. Despite high adoption rates, only 23% of organizations report that AI is exceeding their expectations.

That chatbot era, however, is drawing to a close. The next generation of AI in finance will be defined by tools equipped with the context, system connectivity, and governance necessary to serve as active participants in financial operations — what the broader industry has begun calling "agentic AI," a shift major technology providers including Google, Microsoft, and Salesforce are also investing in.

Standalone AI Falls Short in Finance

For most finance teams, initial AI adoption meant using chat interfaces built on frontier models such as ChatGPT and Claude. Data was exported from existing technology stacks and pasted into a chatbot. In parallel, software vendors rushed to bolt AI features onto legacy platforms, while startups promoted "AI-native" experiences.

Models like ChatGPT and Claude can undoubtedly improve productivity for tasks such as document summarization and other straightforward activities. But that level of utility falls well short of the transformation finance leaders envisioned when making their initial AI investments.

The core limitation of standalone AI is that it exists outside the systems where finance work is conducted. It cannot access accounts payable (AP) workflows, vendor records, invoice data, or payment histories — the very information that provides the critical context needed for meaningful recommendations. Furthermore, because standalone AI operates in isolation from other tools in the finance technology stack, it cannot act on the recommendations it generates. One AI tool might analyze a discrepancy but cannot investigate it; another might recommend an approval but cannot route the invoice to the appropriate person at the right time.

Open Protocols Like MCP Represent the Next Era

Open protocols such as the Model Context Protocol (MCP) are beginning to bridge this gap. MCP, introduced by Anthropic in late 2024 as an open-source standard, provides a standardized method for AI to securely connect to the systems where finance work occurs. Rather than functioning as a standalone chatbot, AI equipped through MCP can access relevant data, understand business context, and perform approved actions in real time.

This represents a significant shift. Where AI was once limited to answering questions, open protocols enable it to become an active participant in financial processes.

Consider a scenario in which a supplier submits an invoice to a buyer's AP team and the price does not match the purchase order. A standalone AI tool could explain the discrepancy or suggest next steps. AI connected through an open protocol like MCP, however, can take a direct role in investigation and resolution: it can identify the correct purchase order and vendor history, compare current pricing against past invoices, pinpoint the likely cause of the mismatch, route the invoice to the appropriate approver, notify relevant stakeholders, and record the resolution — all within the organization's established permissions, approval policies, and workflows. Manual effort decreases, but the team retains full control.

The Impact Is Especially Pronounced in Finance and AP

Each day, finance teams generate and manage enormous volumes of operational data — invoices, purchase orders, vendor records, payment histories, contracts, and approval workflows. Together, this data can provide the context AI needs to make better decisions and take impactful action. At most organizations, however, financial data is scattered across disconnected systems and governed by strict access controls.

Open protocols like MCP provide the structure for AI to connect to financial data securely. Teams no longer need to manually transfer data between disconnected tools; AI can access the right information at the right time while remaining within the guardrails of existing permissions, approval policies, and audit requirements.

The benefits are particularly evident in accounts payable, where a lack of context frequently leads to late vendor payments, undetected fraud, and overspending. Resolving an invoice discrepancy, for example, may require locating the correct purchase order, verifying historical pricing, confirming that goods were received, and identifying the appropriate approver. Navigating a vendor dispute might involve searching through extensive email chains, payment records, and supporting documentation. These tasks are time-consuming, and delayed resolution can create genuine business risk.

With open protocols like MCP, investigations that once required significant manual labor can often be completed in minutes. Issues are resolved faster, freeing teams to focus on higher-value work. In this context, AI evolves from a productivity convenience into foundational finance infrastructure, enabling organizations to make better and more timely decisions without sacrificing accuracy, governance, or control.

Open Protocols Introduce Governance Where There Was None

As AI becomes more deeply embedded in business operations, many finance leaders remain cautious — and security is typically their foremost concern. Financial systems contain highly sensitive information, and any technology that interacts with that data must meet the highest security standards. Finance functions also operate under strict regulatory regimes — Sarbanes-Oxley (SOX) controls in the U.S., data privacy frameworks such as GDPR in Europe, and PCI-DSS requirements for payment data — making governed, auditable AI access a practical necessity rather than a luxury.

Open protocols do not equate to unrestricted access to financial systems. On the contrary, protocols like MCP establish secure, role-based connections between AI and finance systems that respect existing user permissions, approval policies, and audit requirements. Every interaction can be scoped, monitored, and logged.

This approach is often significantly more secure than current practices at many organizations, where employees routinely feed invoices, contracts, and other sensitive financial data into public AI tools. These informal, ungoverned workflows expose organizations to security and compliance risks. Open protocols offer a safer alternative: instead of employees working around existing systems, organizations can provide AI with secure, auditable access to the right information while maintaining the governance, accountability, and control that finance teams require.

What Comes Next for Finance AI

The chatbot era of AI in finance is yielding to a new generation in which open protocols equip AI with the context and connectivity needed to deliver tangible business value. Finance teams that adopt this approach early stand to gain significant efficiency improvements and a durable competitive advantage.

The longer these protocols are in place, the more contextual data AI can learn from, enhancing its ability to streamline routine tasks, surface insights, and support decision-making. Organizations that delay adoption risk falling behind early movers.

At the same time, open protocols are reshaping the role of finance itself. When an organization's financial operating system can manage routine work in the background, teams gain more capacity for higher-value analysis and decisions that genuinely require human judgment. This financial operating system is not defined by any single tool or platform; it represents the direction the industry is heading — one in which AI possesses the context, connectivity, and governance to function as an active participant in everyday financial operations.