BILL launches agentic AI tools to help SMBs manage finance data

Bridging the trust gap

Many small and medium-sized businesses (SMBs) are eager to use artificial intelligence but remain wary of its accuracy. According to research conducted by BILL, nearly 90% of US accounting firms describe themselves as ambitious about using AI technology.

However, the same study found that only about half of those firms trust the accuracy of AI-generated data to any degree. This hesitation stems from a desire for human oversight. Nearly nine in ten finance teams reported that they want centralized or in-workflow control over how AI operates within their specific business processes.

To address this, BILL is introducing agentic AI capabilities. Agentic AI refers to systems that can perform tasks and make decisions to achieve a goal, rather than just following a single prompt. The new tools are designed to give finance leaders the ability to decide exactly where and how these automated systems work.

Risk intelligence and oversight

One of the primary goals of the new update is to protect companies from simple mistakes. In fiscal year 2026, BILL identified approximately $1.5 billion in customer payments that resulted from errors, such as a vendor sending an invoice for the wrong amount or other unusual activity.

The new risk intelligence feature works by checking invoices before they are scheduled for payment. It compares the current invoice against a vendor's established billing history. If the amount deviates from the norm, the system flags it for review during the approval stage. Between August and September 2026, this feature detected an average of $138,000 in mistakes or unusual payments per customer.

To ensure humans remain in the loop, BILL created a consolidated hub for managing AI. This dashboard allows users to monitor every action the AI takes. It also provides the ability to switch specific AI tools on or off or set the specific level of autonomy the AI is allowed to exercise on the company's behalf.

Chat and automated onboarding

The platform is also introducing a natural language chat feature, currently in beta. This allows finance teams to interact with their data by asking questions in plain English. Users can ask about which vendors currently need payment, the status of a specific invoice, what is driving their credit card spend, or how the month-end close is progressing.

Beyond data querying, the company is using AI to streamline the initial setup of the platform. The AI-assisted onboarding tool personalizes business profiles and recommends specific products. It can also help users add vendors and assign default approvers. These features can cut platform setup time by up to 50% and vendor customization time by up to 70%.

The system also includes automatic document labeling for incoming invoices and W-9 forms. While the AI handles the heavy lifting of identifying and organizing these documents, the platform ensures that users still review all changes before they are applied to the account. This balance of automation and human verification is intended to help SMBs adopt AI without sacrificing financial security.

Scaling with network data

The effectiveness of these tools is backed by a massive amount of historical data. BILL's AI is built on more than $2 trillion in payment volume. This gives the system exposure to a vast range of payment patterns, vendor behaviors, and fraud signals.

The company is also making its data more accessible to external tools. Through a Model Context Protocol (MCP) server, the more than 9 million members of the BILL network can connect their accounts to outside AI assistants. This allows them to query their specific BILL data from other platforms. This move follows a trend of increasing developer interest, as the use of BILL's APIs has grown sevenfold over the last two years.

By providing these controls and data access points, BILL aims to move away from fully automated financial decisions toward a model of supervised automation. As René Lacerte, CEO and Founder of BILL, stated, "finance leaders should be able to decide how and where AI works for them."