Qlik Adds Semantic Models to Data Products, With Ossie Interop Still in Development
The capability connects glossary terms to measures and dimensions across analytics and AI. Its Apache Ossie interoperability is an intention, and the release’s own availability language is unresolved.
Qlik introduced Semantic Models on 29 October, a capability for defining, governing and reusing business definitions inside data products. The premise is one every analytics team recognises: a number can be arithmetically correct and still answer the wrong question. Qlik’s chief technology officer, Sam Pierson, states it directly in the release, and the capability is his answer to it, with definitions, expressions and relationships made explicit and reusable rather than rebuilt in each application.
What it does
Semantic Models sit on top of Qlik’s existing data-product capabilities, which already carry curated datasets, ownership, lineage and data-quality information. The addition is business logic: a governed definition of a metric that analytics and AI both read. That combination is the argument worth noting. Curated datasets, ownership, lineage and quality information describe where data came from and whether it is trustworthy; business definitions describe what it means. Qlik’s position is that the two belong in the same object rather than in separate catalogues, and the position is reasonable, because a lineage record tracing a number back to a table without recording which definition produced it is only half a lineage record. The release describes four components.
The first connects terminology from the business glossary to the measures and dimensions used across applications, so a term defined once carries through. The second extracts semantic definitions from existing Qlik analytics applications into reusable data products, which matters for an installed base with years of modelling invested in those applications; it entered public preview on 28 October with general availability planned later in the fourth quarter. The third extends semantic model support to Qlik Answers and to MCP tools, targeted for general availability or shortly after, alongside planned AI assistance for creating and maintaining models. The fourth is interoperability.
The Ossie question
That fourth component deserves scrutiny. Qlik says it is developing interoperability with Apache Ossie, the open semantic interchange specification now incubating at the Apache Software Foundation, with a public preview planned as work progresses. That is a statement of intent rather than a shipped feature, and the phrasing, developing and planned and as work progresses, is deliberate.
Ossie itself is substantial. It defines a YAML format for exchanging semantic metadata across semantic models and ontologies, with reference converters for several vendor formats and production use at Snowflake, Salesforce, Dremio and dbt Labs. What Qlik has announced is an intention to interoperate with it, not an implementation of it, and the release gives no date.
The availability anomaly
One detail is worth flagging. The document carries a 29 October date and states a general availability target of 28 October 2026 for Semantic Models. Either the capability shipped the day before the announcement and the language is stale, or it has not shipped and the stated date has passed. The release does not resolve which.
That matters practically. A semantic model becomes useful once teams can build on it, and a capability whose stated target date has already passed leaves a reader unable to tell whether they can adopt it today.
The customer
The release carries one named customer. Henri Rufin, head of data and analytics at Radiall, says the company can bring its business definitions, metrics and relationships together in a governed, reusable way rather than recreating context for every new use case. It is a sensible endorsement and it describes the intended benefit rather than a measured one; no figures accompany it, and no outcome is reported.
What to watch
Three markers. The first is whether the October general availability date holds, since the release’s own language does not confirm it. The second is whether the extraction capability leaves preview on schedule in the fourth quarter, because that is what makes the offering useful to Qlik’s existing customers rather than only to new ones. The third is whether the Ossie work produces a public preview with a date, since interoperability announced without a timeline is a position rather than a product.