PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteEnterprise semantics is the work of giving business concepts—such as “customer,” “customer segment” and “net revenue”—shared, explicit meaning, then connecting those concepts to the data that represents them. Suresh Srinivas argues that AI agents could make this long-standing goal more practical by combining data metadata, formal business relationships and persistent organizational knowledge. That is a promising approach, not proof that agents can yet answer every business question reliably.
What enterprise semantics means in practice
Imagine asking how many customers a company serves in Europe, or asking for revenue by customer segment. The data may already exist, but different teams might count customers differently, disagree about which revenue measure applies, or use “Europe” to mean different regions. Those disagreements can delay an answer even when the organization has no shortage of data.
As an Amazon Associate I earn from qualifying purchases.
Enterprise semantics aims to make those meanings explicit and connect them to the relevant data. The goal is not simply to load databases into a large language model. As Suresh Srinivas puts it, “LLMs still need to be told what the structured data means, how business concepts are defined, and which data is authoritative.” He is co-founder and CEO of Collate and the OpenMetadata open-source project, and previously worked as chief architect of Uber’s data platform. His article is an industry practitioner’s argument, not an independent assessment of AI data agents.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesWhy glossaries and earlier ontology projects fell short
Glossaries clarify terms, but may not express relationships
A data catalog glossary can help people use a metric name consistently. But a definition in prose alone may not tell a machine which entity a term describes, what properties it has, how it relates to other concepts, or which rules apply. An agent asked for revenue by customer segment needs more than a sentence defining “net revenue”: it needs usable connections between the business concept, the relevant data and the organization’s calculation rules.
#1 Best Overall
Ontology work was difficult to scale and maintain
Earlier Semantic Web efforts developed standards such as RDF, OWL and SKOS to represent machine-readable meaning. Srinivas’s account is that the vision was compelling but enterprise implementation was expensive: ontology work often required scarce people with both technical and business expertise, lengthy workshops and continuing manual upkeep. When systems or business definitions changed, the semantic model could fall behind.
In his words, “The Semantic Web had the right vision and the wrong tools.” That is his assessment of the historical problem, not a claim that the standards themselves have disappeared or that every organization faced the same obstacles.
Rank #2
The three kinds of context in the proposed approach
Srinivas describes an AI-ready context layer as three complementary forms of knowledge. They solve different problems; none substitutes for the others.
| Context type | What it describes | Why an agent needs it |
|---|---|---|
| Data context | Structural schemas, data-quality signals, lineage and usage information | Helps identify what data exists and assess which sources may be appropriate. |
| Semantic context | Formal ontologies, business relationships and rules—for example, how “customer” relates to “net revenue” | Explains what business concepts mean and how to interpret them consistently. |
| Memory context | A shared, persistent record of corrections, feedback and organizational knowledge | Lets agents reuse expert guidance rather than losing it between questions or projects. |
In practical terms, a data agent needs to discover candidate tables, understand which business definition applies and retain relevant corrections from people who know the domain. A glossary may help establish terminology, while relationships and rules make that terminology more actionable for machine reasoning.
What AI could change—and what still needs people
The proposed change is that AI could assist with the costly work around context: populating technical metadata, drafting ontologies for experts to review, and detecting drift when data or business definitions change. If those tasks become easier, organizations may be able to create and maintain semantic models more continuously instead of treating them as large, occasional projects.
That does not establish that the work is autonomous or reliably solved across organizations. Drafted definitions and detected changes still need review; teams must decide which source is authoritative, approve business rules and govern updates. The article presents AI as a way to help people build and maintain context, not as a reason to remove governance.
How strong is the evidence for better answers and lower costs?
The figures Srinivas cites are not independent benchmarks. He describes “seven times more accurate answers” and “86% lower query workloads” as results from “our internal tests.” The article does not provide the test design, sample size, baseline or independent replication, so those results should be read as company-reported internal findings—not as expected outcomes for other organizations.
The article also reports a Gartner forecast that organizations prioritizing semantics in AI-ready data could have 60% lower AI costs by 2027. That is a prediction reported in the article, not a measured result, and the underlying Gartner report is not linked there. It should not be treated as a verified savings guarantee.
The broader argument—that clearer definitions and better connections between data and business concepts can help an agent interpret questions—is plausible. Whether a particular context layer improves answer quality, reduces query workloads or lowers total cost depends on implementation and measurement. The article does not establish universal performance or independently validate its figures.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How to assess an enterprise-semantic approach
Because the article compares no vendors or products, it supports evaluation criteria rather than a product recommendation. Organizations considering this approach can examine:
- Relationships, not just definitions: Can the system represent entities, properties, business relationships and calculation rules, or does it mainly store glossary text?
- Data coverage: Does it bring together schemas, quality signals, lineage and usage information needed to identify and assess relevant sources?
- Persistent corrections: Can approved feedback and organizational knowledge be retained and reused appropriately?
- Maintenance: How are changes to data and business definitions detected, reviewed and reflected in context?
- Governance: Can subject-matter experts approve definitions, rules and authoritative sources, and can teams manage changes?
- Measured outcomes: Are answer quality, workload and total cost evaluated against a clear baseline, with enough detail to judge whether results apply to your own environment?
Srinivas concludes that “The bottleneck that kept this dream out of reach for three decades is gone, and organizational knowledge can now build on itself instead of decaying between projects.” That captures the article’s optimism. Whether the bottleneck has truly gone depends on how well AI-assisted context creation works alongside expert review, reliable metadata and ongoing governance.
Free tools Windows power users keep installed
One-click scans. No signup required.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
