Disclosure: The article described here is sponsored content. MIT Technology Review states it was produced by “Insights,” the publication’s custom content arm, and was not written by MIT Technology Review’s editorial staff. SAP is the sponsoring partner. This article reports what that sponsored piece argues; readers should weigh the framing accordingly.
MIT Technology Review published a sponsored piece, produced in partnership with SAP, arguing that the core failure mode in enterprise AI is not model performance but missing data context. The primary source is Irfan Khan, SAP’s president and chief product officer of Data and Analytics.
Khan is quoted as saying: “AI is incredibly good at producing results. It moves fast, but without context it can’t exercise good judgment, and good judgment is what creates a return on investment for the business.”
The sponsored piece argues that traditional enterprise data strategies strip context when they aggregate information into warehouses, lakes, and dashboards. Metrics survive the move; meaning — such as which customers are strategic accounts, what tradeoffs are acceptable during supply shortages, which contractual obligations apply — does not. Khan’s term for preserving this meaning is a “data fabric.”
The context argument
The sponsored piece uses a supply-chain disruption scenario to illustrate the argument: two systems analyzing the same inventory levels and lead times can reach different conclusions if one has access to business policies, customer priority rules, and extended supply chain status while the other does not. Both move fast. Only one moves in the right direction, the piece argues.
The piece asserts this matters more as AI systems shift from surfacing insights to taking autonomous action. In an agentic workflow, a system that optimizes for the wrong objective does not just produce a bad report — it executes a bad decision.
The sponsored piece cites survey data from three sources. A McKinsey survey is cited for the claim that by the end of 2025 half of companies were using AI in at least three business functions. A Capgemini report is cited for the claim that only 9% of organizations felt fully prepared to integrate and interoperate across their data systems. A BARC survey is cited for the claim that more than two thirds of enterprises with data fabrics saw improved data accessibility. A fourth claim — that only one in five organizations consider their data approach highly mature — is cited without a named source.
What a data fabric means in this framing
The sponsored piece defines a data fabric as an abstraction layer spanning infrastructure, architecture, and logical organization — distinct from simply consolidating data into a single repository. Key capabilities described include federation across multiple environments rather than forced consolidation, a semantic or knowledge layer often supported by knowledge graphs and catalog-driven metadata, and governance and policy enforcement operating across the fabric.
For agentic AI specifically, the piece argues that knowledge graphs allow natural-language queries grounded in business logic rather than database schemas. Without a common knowledge layer, the piece argues, multi-agent coordination breaks down when different agents work from different slices of the same underlying data.
Khan is quoted in the sponsored piece as saying: “It empowers confident, consistent decisions, and when these elements all come together, AI just doesn’t analyze and interpret the data — it drives smarter, faster decisions that really create business impact.”
The framing throughout the sponsored piece aligns with SAP’s product positioning in the enterprise data and analytics market.