Why Asset Managers Need a Unified Data Foundation Before Scaling AI

In February 2026, Mercer's survey report revealed that 91% of asset managers planned to increase their use of AI in the next 12 months. Since we are now deep into 2026, it is safe to say that managers are in the thick of piloting, implementing, and exploring AI use case integrations. Nearly 7 in 10 reported data constraints as a significant barrier that prevents further AI adoption in their investment process.ⁱ Hopefully, solving those data constraints is top priority for firms. Otherwise, they will be rolling the boulder up an infinite hill.

AI is only as powerful as the data infrastructure it runs on. This technology is not a magic solution that can independently fix broken operational processes. AI is an amplifier that magnifies the quality — or lack thereof — of the underlying data infrastructure. Managers need to scale AUM, not data errors. No matter how ingenious the AI tool is, it will hold and propagate bad information if fed from unreliable sources. Here is why sophisticated AI model intelligence will not be the differentiator; and why a unified data foundation will be the true differentiator for asset managers.

AI-ready data foundation for governance and auditability

According to a report released in August of 2026, 81% of senior technology leaders in UK asset and wealth management say poor data and legacy systems are limiting their AI progress, and 44% say their data does not have the quality, lineage, and traceability needed for regulatory confidence.ⁱⁱ Of course, technology leaders don’t really need to see these statistics; they are living them. There is considerable pressure to move swiftly and demonstrate value. But speed to market should not sacrifice governance and precision. Some teams have rushed ahead running AI pilot projects in environments separate from core workflows. Some are running them in core workflows, but without a unified data layer. These approaches can be counterproductive. Think of the AI-ready data layer as a mandatory prerequisite to agentic adoption, enabling governance, auditability, and explainability.

We’re talking about constructing the right foundation that supports the AI building, no matter how many floors are added. By definition, foundations don’t go on top. Some CTOs have approached operationalizing AI by adding centralized investment data and a trusted system of record on top of AI workflows. They open the firms up not only to underperforming initiatives but also costly AI risks. Only 22% of asset managers said they are very confident they could pass an independent AI governance audit.ⁱⁱⁱ What is the issue with this approach? The big data problem.

The big data problem and the operational risks that follow

For years, we have emphasized the critical importance of a centralized, golden source of data to rectify the persistent fragmented data problems that arose with the proliferation of closed SaaS tools jammed into tech stacks. These point solutions were a necessary step in modernization to be sure but produced an industry-wide data problem. This data problem naturally extends to agentic AI workflows.

AI agents, like their human counterparts, must be able to use the firm's disparate data from these tools to execute functions like reconciliation, reporting, and accounting. That means the underlying data infrastructure must be able to automate data flows to normalize and consolidate structured and unstructured datasets that pass through several systems before it reaches a report or analytic output. Deploying AI on top of fragmented, misaligned copies of data, or un-reconciled data stacks creates severe liabilities.

Autonomous AI needs a trusted system of record

Let's say treasury and investment ops agents are tasked with autonomously managing treasury cash balances, routing margin payments, and booking trades across multiple prime brokers and custodians. The agent connects to multiple disjointed legacy systems via external APIs, pulling independent copies of position and cash data. Lacking a shared data layer, data drift occurs across the firm's systems because data moving through the trade lifecycle flows through pipelines from external counterparties, custodians, pricing providers, through numerous internal front-, middle-, and back-office systems. Each system typically has its own data models, logic, and its own treatment of lifecycle events.

Subsequently, the AI agent may treat a pending corporate action and an intraday trade rebook inconsistently. The firm now has a false view of available cash and exposure, leading to incorrect margin payments, duplicated trades, overdrafts, or failed settlements. Imagine if the agent then programmatically authorizes a high-value transaction or cash transfer, triggering failed trades and a severe settlement break. This would damage if not destroy counterparty confidence and create immediate liquidity strain.

Every agent must reason over the exact same data source that governs the rest of the firm's operations. Firms that have worked hard to solve this data problem for their traditional operational systems will possess the key to solve the data problem for AI deployments.

AI-ready data foundation gives autonomous agents embedded governance

AI agents are subject to the same monitoring, review, and documentation as traditional models and tools. CTOs should deploy their AI agents on top of a unified data foundation and infrastructure that provide embedded governance. Clear agent accountability ensures transparency, compliance, and reliable performance across AI-driven investment workflows. The EU's AI Act, GDPR, US regulations in states like Colorado, California, and New York, and evolving federal frameworks are the drivers of external accountability.^iv Regardless of the policy frameworks, internal governance is crucial. The operational risk in using AI in investment management is too imposing. Plus, there are additional risks to consider when deploying pilots in fragmented data environments without governance architecture. Firms cannot afford to report bad data to regulatory bodies or investors, leading to lawsuits, default risks, and cash flow shortages, and failure to deliver on the mandates that the fund was set out to do.