OneStream Deepens AI Strategy, Launches Interoperable Agents
Financial software vendor OneStream announced on Tuesday the expansion of its agentic AI capabilities with the launch of a financial agent layer, enabling interoperability with third-party systems such as OpenAI ChatGPT, Anthropic Claude, and Microsoft Copilot. It also introduced predictive agents that support natural language queries. This move follows SAP's announcement last week of its "autonomous enterprise" initiative, and industry analysts believe agent interoperability has become a standard requirement. OneStream emphasizes security governance and audit trails to ensure that external AI tools accessing data comply with financial regulatory requirements.

At a Glance
- Financial software maker OneStream announced on Tuesday newagentic AIcapabilities, including a feature that supports interoperability with third-party systems such as OpenAI's ChatGPT, Anthropic's Claude, and Microsoft's Copilot.
- At the core of the release is what OneStream calls a "financial agent layer," designed to let its AI agents work with external tools.
- "We're starting to see more and more customers say, 'I'm fully using ChatGPT or Claude or Microsoft Copilot,' which is not surprising," OneStream data scientist and software engineer Andrew Shea said in an interview. "We responded to that quickly."
Deep Dive
Enterprise software vendors are accelerating the embedding of agentic AI into corporate finance and other business functions, raising broader questions about how governance, auditability, and interoperability play out at scale.
Last week,SAP launched a new AI initiative called the "autonomous enterprise". As part of that initiative, the company said it will enable agent-to-agent interoperability with external applications.
A Constellation Research report describedSAP's move as a "table stakes", noting that "all agents must use common protocols, share data, and be interoperable."
OneStream's latest release comes about a year after it launcheda suite of AI agents(including a "financial analyst agent" and a "deep analysis agent"). On Tuesday, the company added a "forecast agent," allowing users to ask questions about forecasts in natural language.
Previously, OneStream focused on deploying "native" AI agents that run only within its own platform. The company's new approach aims to let these capabilities work with external systems while retaining the controls finance teams need.
In Tuesday's press release, the company said its new financial agent layer uses the so-called open Model Context Protocol to provide any AI tool with "secure, governed access to OneStream data, with the logic, auditability, permissions, and audit trails finance requires."
"The key is making sure OneStream's financial intelligence layer can be present wherever customers are working," Shea said.
He described MCP as "the USB interface for AI systems to talk to each other," enabling organizations to connect OneStream's agent tools with any external AI ecosystem.
"When ChatGPT is asked to provide a dashboard showing trends, variances, and forecast predictions, it retrieves reports from OneStream through the financial agent layer and generates executive-level output in minutes," the company said in the press release.
Every interaction through the OneStream financial agent layer is authenticated against the user's OneStream identity and "enforced based on their existing role-based permissions," ensuring that users working in external AI tools can only access data they are authorized to see, the press release said.
The company added that administrators can retain visibility into system usage and activity within OneStream, "providing the control and auditability finance operations require."
"If numbers aren't based on a trusted source of truth, you end up with wrong numbers and trust erodes quickly," Shea said.