AI has already transformed analysis, dramatically increasing its speed and lowering its cost. Questions that once took analysts days to answer can now be resolved in minutes, with dashboards, forecasts, and summaries generated almost instantly. But more analysis does not necessarily make decisions easier. Executives must sort through a growing volume of outputs to determine what is reliable and which changes in operations and market conditions have meaningful financial implications.
The result is a growing mismatch: AI can produce analysis almost instantly, while people still need time and context to make sound decisions. The answer is not to generate more or faster analysis but to give decision-makers the right information at the right time, along with the context they need to understand the choices and their implications. This requires a system that can distinguish important developments from the much larger volume of information that requires no action.
The good news is that such a system is beginning to emerge. We call it the “always-on finance intelligence layer,” or, more succinctly, the intelligence layer. It combines trusted company data and business context with AI to show company leaders how financial performance is linked to operational and market metrics. This system helps leaders see not only whether revenue, margin, cash flow, or other outcomes have changed but precisely why they changed, which factors caused the change, and where action is needed. These insights can inform both day-to-day operational decisions, such as how to address emerging customer retention or supplier risks, and longer-term strategic choices, such as where to invest capital or how to reshape a company’s portfolio of businesses.
Much of this analysis is already possible with existing tools, but typically someone must recognize the need for it and initiate the work. An intelligence layer operates continuously in the background, bringing emerging risks and opportunities to decision-makers’ attention without requiring them to initiate an analysis.
By giving leaders access to the financial insight they need when they need it, an intelligence layer can bring the financial consequences of decisions into view before decisions are made. Building an intelligence layer, however, requires more than new technology. It requires the CFO to rethink how financial insight is created, delivered, and used across the company.
CFOs can lead the way
The CFO is well positioned to lead the development of an intelligence layer because finance connects decisions across the business to their financial consequences. Finance can identify which operational and market signals have the most impact on company performance, helping leaders make choices about day-to-day operations, resource allocation, and long-term strategy.
Finance has an enterprise-wide view
The CEO can set the agenda of decisions that matter, while the chief information officer builds and runs the technology. But CFOs are enterprise-wide leaders who are not beholden to any single business unit and who bring common finance language to strategic choices and value creation opportunities. They can work with their peers across functions to connect operational and market data to financial outcomes. Thus, finance is ideally suited to define the financial and operational signals the intelligence layer should track, establish common definitions for the metrics it uses, and assess the financial implications of changes in those indicators. Realizing the full potential of an intelligence layer requires close collaboration among finance, business, and technology leaders.
The CFO’s first task is to choose the decision to support
Before the intelligence layer is built, the CFO can work with business leaders to identify the significant decisions it needs to support, such as pricing, inventory, hiring, or investment. For each one, they can ask four questions:
- Who makes the decision?
- Where and when is it made?
- What financial consequences do leaders need to understand before making it?
- Who is accountable for the result?
The answers help determine what information the intelligence layer needs to provide and where it should appear. If a pricing decision is made within a particular workflow, for example, the relevant financial insight should be available there when the decision is made, rather than delivered later in a separate report. Once these requirements are clear, the company can begin building the data and technology that supplies them.
What is an intelligence layer?
An intelligence layer is a system that draws on trusted financial, operational, and external market data from core enterprise systems and public sources. Powered by AI (though not built on a specific model, because the new ones arrive too quickly), it combines this data with shared business definitions and the context leaders need to interpret performance, including plans, historical performance, external benchmarks, and business rules.
Its output can show what changed and why, what may happen next, and what actions leaders might take, customized to the role and the decision at hand. For example, frontline employees can see operational issues they can act on, business unit leaders can see performance drivers, and CEOs and CFOs can see the signals most pertinent to enterprise value and strategy. The intelligence layer can deliver these insights through whatever tools employees already use, including dashboards, natural-language queries, AI agents, or applications such as planning and customer-relationship-management (CRM) systems. AI agents can then organize the information for a particular decision, flag issues that require attention, or suggest possible actions. Executives remain responsible for interpreting the information and deciding what to do.
Dashboards continue to be a useful way to view performance. But unlike a proliferation of dashboards with a wall of KPIs that often leaves senior executives unclear on what is signal and what is noise, an intelligence layer creates a shared, trusted view of performance. The latest models of intelligence layers can serve as excellent sparring partners for stress-testing decisions and brainstorming ideas. Rather than just handing back an answer, they can conduct thoughtful conversations with leaders, surfacing relevant points from the vast body of knowledge they can access. Just as important, they remove the handoff between finance and the business by embedding the same information in the tools different parts of the company use to make decisions.
An illustrative example: A B2B software company spots risks early
Companies have long used predictive models to identify customers at risk of leaving. But developing and deploying models traditionally requires substantial technical expertise and investment. Newer AI tools can make it less costly to identify and interpret emerging signals, and the intelligence layer can make it easier to connect them to financial outcomes and recommend appropriate responses.
Imagine how a hypothetical B2B software company could use an intelligence layer to spot early signs that customer renewals may be at risk. The intelligence layer could continuously monitor a set of leading operational indicators, including a decline in a customer’s admin logins, lower use of licensed seats, or unresolved service issues. The system could use this data to identify patterns that often emerge two or three quarters before a customer decides not to renew.
The intelligence layer could flag the accounts that pose the greatest risk to revenue forecasts. It could also quantify the likely financial exposure and recommend actions for customer support or commercial teams to take. For accounts with lower product adoption, for example, recommendations could include a review of product fit, targeted technical support, or executive outreach.
How to build an effective intelligence layer
Building an effective intelligence layer requires companies to put the technical and organizational foundation in place, build credibility around the data the system uses, and embed the tool into how leaders make decisions.
Change is both organizational and technological
A well-built intelligence layer consists of five building blocks. The first is organizational and, in our experience, often the most neglected; the other four are technical.
- An operating model for using the intelligence layer. Companies need to determine when and where insights will be used and who is responsible for acting on them. This might require weekly operating reviews, monthly business reviews, or planning cycles. They also need clear owners for the metrics the intelligence layer tracks. Without these arrangements, the technology may produce useful insights without changing how decisions are made.
- A shared business language that defines important metrics, specifies how they are calculated, and identifies the authoritative data source for each one, so that finance, sales, and operations share the same understanding of the numbers. Some organizations begin with a registry of business terms as they are used within the company and gradually develop it into a more sophisticated semantic layer as the system grows. Shared definitions are what make the intelligence layer’s outputs credible across business units; without them, each function may trust only its own figures. Establishing this credibility is difficult, as we discuss below.
- Connectors link the intelligence layer to enterprise and external data sources, including enterprise resource planning, CRM, planning, and operational systems, as well as feeds for competitor benchmarks, news, and market data. Before establishing these connections, companies will need to inventory their data sources, determine where critical information resides, and identify systems or data in need of modernization or consolidation. They can then prioritize the connections needed to support the most important business decisions. Existing technologies, including cloud storage, APIs, and standards such as the Model Context Protocol, can help the intelligence layer securely access and use this information.
- Skills are reusable AI capabilities that perform specific tasks, such as identifying the drivers of a variance, comparing performance with peers, detecting an emerging trend, or modeling the financial implications of a scenario. What’s important here is not quantity but quality; hundreds of bespoke skills are unnecessary, but each has to be expertly written. A small library of skills can be reused across different decisions and functions. For example, a financial-services company defined roughly 20 core skills, three-quarters of which could also be used outside finance. The same skill for identifying variance drivers could help a CFO understand why revenue is below plan, a commercial leader assess pricing performance, or an operations leader review service levels. Because these skills use the same definitions, calculations, and data sources established by the shared business language, their analysis remains consistent across the company.
- An orchestration layer coordinates the other components of the intelligence layer. When a user asks a question, it determines what data and skills are needed, brings them together, and delivers the result as an analysis, chart, or recommendation. The same underlying data, definitions, and analytical capabilities can therefore support dashboards, AI agents, and other applications across the company. Every result remains traceable to its source and is limited to information the user is authorized to access.
Build company-wide credibility
An intelligence layer can succeed only if leaders trust the facts it presents and the recommendations it makes. Therefore, it’s crucial to start with a shared business language consisting of the most important metrics, how they are defined and calculated, and the authoritative source for each. Establishing this foundation is difficult because in many organizations, business units and functions do not always understand or accept data produced elsewhere in the company. Skepticism can be strongest when the numbers affect the narrative around performance assessment or how resources are allocated.
Business units may still retain views tailored to their operational needs. But when senior leaders convene to assess enterprise performance, allocate resources, or resolve cross-business trade-offs, they need to work from the same standardized definitions and underlying data. Without this common foundation, discussions can quickly become debates over whose numbers are correct rather than what actions the company should take.
Companies can build credibility into the intelligence layer from the beginning. They can appoint clear business owners for critical metrics and create cross-functional forums to surface and resolve disagreements about definitions and calculations. This is primarily a leadership challenge, because business units may need to give up their own versions of disputed numbers and agree on the definitions and data the company will use. Once those disagreements are resolved, the agreed-upon results can be documented in a shared registry and used by the intelligence layer.
Embed the intelligence layer into strategic and operational decision-making
Financial results are lagging indicators: They show what has already happened. Operational and market data can provide early indications of where financial performance is headed, but identifying these indications and understanding their financial significance can require extensive analysis. An intelligence layer can continuously examine a wide range of variables, identify changes that might otherwise go unnoticed, and help leaders understand their potential financial implications while there is still time to act. This capability can inform strategic decisions as market conditions evolve and help companies detect operational problems before they affect financial results.
An illustrative example: An industrial company identifies opportunities to reallocate capital
Imagine how a hypothetical $10 billion diversified industrial company could use an intelligence layer to identify opportunities to reallocate capital as business and market conditions change. The company traditionally reviews capital allocation once a year, potentially missing opportunities between planning cycles. By continuously monitoring business unit performance, customer demand, and market valuations, the intelligence layer could alert executives to opportunities to invest in growing businesses or divest underperforming ones.
For example, suppose orders for the company’s automation products rose 15 percent above plan for two consecutive quarters. The intelligence layer could assess whether the increase reflected sustained customer demand or temporary distributor restocking. If demand appeared durable, it could recommend committing $60 million to expand production capacity, with funding released in stages as demand was confirmed. Executives could then begin planning and approving the expansion sooner, potentially bringing new capacity online months earlier than if they had waited until the next annual review.
At the same time, the intelligence layer might detect a rise in valuations for building-products companies that makes it attractive to sell the company’s lower-return building-products unit. The layer could flag the opportunity for executives, who could then assess whether the estimated $2.2 billion in proceeds from a sale, net of taxes and separation costs, would exceed the roughly $2 billion value of retaining the business. Executives could then decide whether to pursue a divestiture and use the proceeds for acquisitions, share buybacks, or debt reduction.
The intelligence layer can also connect information across functions to surface emerging risks.
An illustrative example: An industrial company receives early warning about a supplier risk
Imagine how an industrial company could use an intelligence layer to identify supplier risks that might otherwise go unnoticed. Its traditional procurement reviews focus only on suppliers with large annual spend, potentially overlooking smaller but critical suppliers.
To help leaders decide when a supplier poses enough risk to warrant action, the company could build an intelligence layer that connects information about its suppliers with the components they provide, the products that depend on those components, and the revenue associated with those products. Procurement, operations, and finance could agree on the definitions and data sources the intelligence layer would use, so that leaders evaluating a supplier work from the same agreed-upon information. The intelligence layer could flag early signs of supplier risk and present that information to users along with the company’s exposure to each supplier.
The intelligence layer might identify a relatively small supplier that would ordinarily escape attention but provides a critical component for a major product line. If that supplier’s lead times began slipping and its financial health deteriorated, the intelligence layer could flag the risk and show leaders the financial implications of not securing that component. Leaders would then decide whether to establish a second source, build inventory, or take another action to mitigate the risk.
Five considerations to guide companies through the process
As this article has described, building an intelligence layer is a complex technological and organizational undertaking. To summarize, companies that follow these five guidelines will be best positioned to succeed:
- Start with important decisions and work backward. Identify the recurring decisions the intelligence layer needs to support, then determine what financial and operational information leaders need to make them. Every KPI should inform a decision; otherwise, it adds noise.
- Put KPIs in context. Give leaders the information they need to interpret each KPI, including performance against plan and over time, relevant benchmarks, underlying business drivers, and potential financial consequences.
- Establish one source of truth. Use agreed-upon definitions and authoritative data sources for important metrics so that leaders work from the same underlying information, even if different users see different views of it.
- Automate the mechanics, not the judgment. Use AI to automate data collection, reporting, and routine analysis, but leave business judgment and decisions to the people running the business.
- Embed, learn, and scale. Use the intelligence layer in recurring business decisions, refine it based on experience, and extend it as leaders gain confidence in the approach.
As AI makes it faster and easier to collect and analyze data, analysis alone is unlikely to provide a competitive advantage. The greater opportunity is to ensure that leaders have the right financial insight when they make important decisions. Are pricing, inventory, hiring, and investment decisions made with their financial consequences already clear? Building an always-on finance intelligence layer that makes the answer yes is the CFO’s next job.
.jpg?cq=50&mw=767&car=16:9&cpy=Center)