Analyst Perspectives

OneStream offers a platform designed to serve the needs of accounting and financial planning and analysis (FP&A) organizations. The software handles financial close and consolidation, planning and budgeting, analysis and reporting.

OneStream recently held its annual user conference, Splash, in Nashville. In attending this meeting, my focus was once again on the progress the company has made over the past year in the areas of predictive artificial intelligence (AI), generative AI (GenAI) and agentic AI. AI technology is important because of its potentially positive impact on organizational performance. This is especially true for the office of finance because the technology lends itself to improving the productivity of the department by automating repetitive work, organizing workflows and improving process execution by eliminating errors at the source.

Agents and agentic AI, in particular, are all the rage now, eclipsing last year’s focus on AI and GenAI. They automate work almost effortlessly so that repetitive and boring tasks get done with the least amount of effort and, perhaps, more consistently. AgenticAI does not replace other forms of AI. Agents are programmatic actors that amplify the capabilities of a software application, often using predictive and generative AI techniques. ISG Research asserts that by 2028, almost all business software providers will have augmented the capabilities of their applications with some agentic AI capabilities to lower costs, upskill users and improve customer service and agility.

The salad days of any major technology breakthrough are almost always at the start when anything conceivable seems possible, and the obvious-in-retrospect issues and unforeseeable gotchas have not yet intruded on happy visions for the future. Knowing this, OneStream and most other providers have been purposely restrained in product roadmaps. This is especially important for accounting and finance because accuracy, reliability and trustworthiness are paramount in this part of any business, so the software it uses must support these objectives and deliver what is promised. With that in mind, OneStream is clear that as it rolls out AI technology, it aims to create functional, reliable systems that address the most common use cases as fast as possible.

It's helped that OneStream had been focusing on machine learning (ML), a foundational element for predictive AI, for years before ChatGPT and generative AI came along. Although the concept of predictive analytics and its use in forecasting and planning has been around for decades, very few enterprises used it in a meaningful way. One barrier has been the lack of analytical talent, while another has been insufficient data on which to train ML models. Because of the focus on all forms of AI, enterprises are beginning to address the latter issue, and OneStream has a data architecture that supports AI.

OneStream’s architecture has been data-centric from the start because data is essential for accurate forecasting, planning and analysis and for streamlining financial consolidation and close motions. Over the past decade, data technologies have advanced to make business data more accessible on application platforms. OneStream’s Analytic Blend offers a user-friendly, governed and semantically optimized data store that supports its AI offerings. It enables instant access to large volumes of integrated financial, operational and external data, eliminating the need for repetitive and time-consuming data preparation and providing the proverbial—and until recently the elusive—“one source of truth.” A major benefit from better data is improved business planning. ISG Research asserts that by 2028, one-fourth of FP&A organizations will implement integrated business planning, bringing together operations and financial planning on a single platform to improve the business value of planning and budgeting.

Because of all of the attention paid to ChatGPT, business software providers have spent the past two-plus years working out how to deliver AI and generative AI technology to users in ways that are trustworthy, reliable and practical. These are being delivered as part of an application’s functionality, as easily customized objects or through studios that enable software partners, implementation partners or trained enterprise specialists to create specialized or complex agents. At the conference, OneStream presented its SensibleAI portfolio, some elements of which are generally available, in limited release or on its 12-month roadmap. The main components include SensibleAI agents, SensibleAI Forecast and a SensibleAI Studio.

SensibleAI agents have financial intelligence designed to analyze, synthesize and act on financial and operational data and documents. These currently come in four flavors:

• Finance Analyst works directly with OneStream’s data store, using natural language to interact with financial data to produce textual answers to questions as well as tables, charts, reports, dashboards, graphs and other visualizations. This substantially reduces the amount of manual effort and user training required, bringing once specialist activities into the hands of individuals in a self-service mode. Even rookie financial analysts can rapidly execute roll-ups and create contingency plans. Budget owners with no formal training can see their variances immediately at period end, business analysts can create and quickly update financial forecasts and executives can interactively drill down into the data to quickly see the numbers behind the numbers.
  
  
• Operations Analyst enables business users to query operational and transactional data and quickly get performance insights, spot anomalies, identify trends and locate inventory, as well as create interactive dashboards, reports and visualizations
  
  
• Search Analyst responds to natural language queries to immediately answer questions related to enterprise content and summarize documents, policies and reports using retrieval augmented generation combined with context to get reliably accurate information.
  
  
• Deep Analysis is designed to surface insights from this proprietary content to find specific information or data, find patterns or spot outliers. Enterprises have massive amounts of unstructured information contained in digital documents, but extracting and gaining an understanding of the content has been a practical impossibility until GenAI became available. The information uncovered can provide insights to support analyses, extract and summarize information to support decision making or bolster situational awareness.

SensibleAI Forecast harnesses the power of predictive analytics to enable anyone to make statistically better forecasts faster. The software applies machine learning techniques to internal and external data to find the best-fit projection at any degree of granularity. By eliminating the need for data preparation and automating routine steps, users can substantially reduce planning cycle times. This, in turn, removes a longstanding barrier to agile planning and the ability of managers to examine contingencies to be better prepared to deal with an uncertain business climate. To ensure trust in the results, the system provides visibility into the factors behind the forecast and the modeling techniques employed in that forecast.

To facilitate the creation and adoption of agentic systems that build on and enhance its core software, OneStream is offering a SensibleAI Studio. This is not aimed at casual users but instead will be used by partners and well-trained specialists in its customer base. The Studio facilitates the creation of custom agents to execute industry or enterprise-specific tasks or to take on specialized roles in business processes. The Studio integrates quantitative and generative AI routines—like machine learning, anomaly detection and natural language to SQL translation—directly into financial workflows. Its no-code Genesis Blocks allow users to build and deploy AI routines for planning, forecasting and reporting. The AI Routines Library provides automation support for AI applications, including software to manage machine learning (including classification, regression and clustering), time-series anomaly detection and cleansing, natural language to SQL query translation as well as data transformation and enrichment. One of the benefits of the Studio is its ability to control agentic design elements such as prompts and queries, which can increase the accuracy and efficiency of the system rather than leaving it to an individual to build these on their own. The bulk of the Studio capabilities covered at the event are either available now or will be in some form of release before the end of 2025.

If history is any guide (and it probably is), the capabilities of business applications like OneStream will become available long before mainstream adoption. Our research has repeatedly found that finance and accounting organizations are technology laggards. However, history also shows that the phrase, "skate to where the puck is going, not where it has been" (attributed to Wayne Gretzky) applies here: Providers must anticipate future needs to achieve market success. Moreover, software providers must also identify and cultivate early adopter customers to ensure success and prove the technology is trustworthy, reliable and practical.

For finance and accounting organizations that are reluctant to be on the bleeding edge of technology, I strongly recommend taking a fast-follower approach to technology adoption. That is, establishing a standing group with clout that stays abreast of relevant technologies and anticipates their introduction into departmental processes, laying the groundwork for change management in the department. Doing so limits risk and facilitates organizational change so an enterprise can improve its performance.

Regards,

Robert Kugel