gshc2020.com

That said, there's growing evidence that organizations need to take action to scale their AI programs. Among the issues is whether the enterprise can supply the data and context that models need.

• McKinsey researchers recently reported that fewer than 10% of companies have fully scaled AI and that more than two-thirds of high-performing companies identify data as the primary obstacle to enabling it.

• KPMG researchers reached a similar conclusion: 58% of executives cited data readiness and access as the biggest challenge to deploying AI agents.

But does that context gap mean the organization is missing technology?

Many vendors would argue the answer is yes. If AI needs more context, then the solution is to buy a new layer that’s “built for the agentic era.” Admittedly, some of these technologies may be useful or even necessary. But for many enterprises, the fastest path to more context might start with an easier question: How much of the context can be delivered by technology my organization already owns?​

In practice, context consists of three things:

1. Semantics Or Business Meaning: business glossaries, data governance tools, master data models, semantic models, policy repositories and data dictionaries.

2. Operational State: master data management platforms, hierarchy management, relationship management, reference data workflows, stewardship workflows, data quality rules and exception management.

3. Traceability: data catalogs, metadata management, lineage tools, stewardship histories, governance workflows, approval records and audit trails.

But context can't be kept accurate, consistent and trustworthy through definitions and metadata alone. It requires an operating model with the people, processes and decision rights that govern how context is created, maintained and changed over time.

Definitions change. Source systems disagree. Policies evolve. Business rules collide. Someone has to decide which definition applies, which source is authoritative, which relationship matters and how exceptions are handled. Those decisions are what allow context to support AI models reliably.

​This is where many organizations get the starting point wrong.

​They already own many of the capabilities required to provide AI with context. Master data platforms contain trusted business entities and relationships. Data governance tools capture ownership, policies and business definitions. Catalogs document metadata and lineage. Workflow engines record approvals, stewardship activities and operational decisions.

Individually, these capabilities work well. They support data governance programs, regulatory initiatives or transformation programs. The problem is that they're rarely designed and later operated as a single enterprise context capability.

As a result, context becomes fragmented. Business definitions live in one platform, stewardship processes in another, lineage somewhere else and operational rules inside applications. Each capability creates value on its own, but AI needs them to work together. Doing the work to understand alignment and gaps might, very well, be the best and most cost-effective place to start.

From Stand-alone Technologies To An AI Context Capability

A key part of this approach isn't just to inventory tools but to determine where business meaning, operational state and traceability live today. Organizations need to understand whether those capabilities are concentrated in a few trusted systems or distributed across multiple platforms, which sources are authoritative, where conflicts exist and how these capabilities can be connected into a common context layer.

Organizations also need to develop a cross-cutting or unified operating model that allows context to be produced, made accessible to AI systems and governed, maintained and improved over time across the distributed tool set. Existing semantic, operational and traceability capabilities need to be connected through common ownership, decision rights, change processes and accountability models. Without that alignment, context remains fragmented across teams, systems and workflows, even if the underlying tools are technically connected.​

​For many enterprises, this could be the fastest and most practical path to AI context. It allows the organization to reuse existing investments, identify the gaps that genuinely require new technology and create a pragmatic approach to expose trusted data and business meaning to AI systems.

The Bottom Line

Enterprises are under growing pressure to demonstrate measurable value from AI. Strengthening the organization’s enterprise data foundation, especially the context layer, is critical to making that value measurable, repeatable and trusted.

That may sound daunting. But the good news is that many enterprises aren't starting from zero. In the organizations furthest along, much of the machinery needed to supply AI with context is already in place: MDM, data governance tools, catalogs, lineage, policies, workflows and operational systems.​

Before investing in another context platform, ask yourself: Have we fully connected the context we already have?

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?