Data AnalyticsData governanceMulti-cloud
Ataccama open-sources Apache Ossie converter for AI trust
SOFIAH NICHOLE SALIVIONews Editor
Ataccama will open-itions and data quality signals to the open semantic specification
The converter is designed to let AI agents and analytics tools understand both what enterprise data means and whether it is reliable enough to use. Apache Ossie began as Snowflake’s Open Semantic Interchange and is now an incubating project within the Apache Software Foundation.
The software will carry semantic definitions and live quality indicators from Ataccama ONE into Ossie. This would allow trust signals to move with metadata across analytics and AI systems, rather than remain tied to a single dashboard or platform.
The development comes as companies try to make AI systems work with business data spread across layer can define terms such as revenue or customer, but those definitions do not show whether the underlying records are complete, current, or failing key checks
Ataccama’s approach is to evaluate quality in platforms, then pass those signals through Ossie-compliant YAML. This means organisations do not need to move all data onto one platform before making quality context available to AI tools
The issue has become more pressing as AI systems move beyond summarising information and begin recommending decisions or taking actions with less human review. In those cases, even a well-defined business metric can produce the wrong result if the data behind it is late, incomplete, or otherwise below an accepted threshold.
Snowflake said the contribution expands the scope of portability in semantic standards beyond definitions alone.
“An open semantic standard promises that enterprises can define their business once and carry that understanding across the tools that use it,” said Josh Klahr, Head of Product Management at Snowflake. “Ataccama is extending that portability to data quality, giving organizations a consistent way to make those signals available across platforms rather than rebuilding them for each environment. That is an important contribution as enterprises connect more of their data to AI.”
Ataccama said the converter will expose several forms of information to AI systems, including warnings when data falls below a quality threshold, continuously updated trust signals based on current checks, and links to further evidence through its MCP Server when an agent needs more detail about a problem.
According to the company, an AI agent could use the immediate signal in the semantic model to decide whether to qualify an answer, investigate further, or avoid acting on questionable data. Through the MCP Server, it could then retrieve information on which quality checks failed and which records were affected.
Ataccama argues that this closes a gap between how people and machines assess business data. In traditional business intelligence workflows, users often catch bad data by comparing it with prior reports, baseline figures, or their own familiarity with the numbers.
Jessica Smith, Chief Product Officer at Ataccama, said AI systems often lack that operational context even when they understand the formal definition of a metric.
“AI exacerbates the impact of incorrect data,” said Smith. “In the BI era, bad data typically got caught because it conflicted with something already known – another report, a trusted baseline, someone’s memory of the last board deck. AI agents need that same level of operational perspective. An agent may know what revenue means, but not whether the revenue data it’s reading is complete or current, and it will still produce a convincing answer. Opening our converter to Apache Ossie will help agents inherit the context that people used to supply, and the critical ability to know whether data should be trusted, thanks to Ataccama’s data quality signals now available in the semantic layer itself.”
Ataccama is open-sourcing the converter to avoid creating another proprietary layer that customers would have to rebuild for each platform. It framed the move as part of a broader industry effort around Apache Ossie, which has drawn contributions from vendors including Snowflake, Databricks, and dbt.
The announcement also reflects a wider shift in enterprise data management, where semantic models are becoming a common way to organise definitions for use across business intelligence, analytics, and AI systems. Attaching quality information directly to those models could make them more useful in environments where software agents are expected to act on data rather than simply display it.
The converter is the latest step in Ataccama’s effort to make enterprise data accessible to AI systems through tools including its MCP Server and ONE AI Agent.
ChatGPT
Key takeawaysExplain why it mattersCreate action planFuture watch
Claude
Key takeawaysExplain why it mattersCreate action planFuture watch
Perplexity
Key takeawaysExplain why it mattersCreate action planFuture watch
Grok
Key takeawaysExplain why it mattersCreate action planFuture watchShareShareAdd us as a preferred source on Google
Image: Jessica Smith and Josh Klahr
Related stories
Cloudera launches hybrid cloud platform for AI workloadsExclusive: Cloudera builds Hadoop-free future platformCloudera named leader in Forrester’s data lakehouse waveDatabricks deepens Microsoft tie-up into the 2030sAI transforming B2B discovery – but human trust endures
Top stories
xAI launches Grok 4.7 for coding & knowledge workGuidePoint adds AI identity services as gaps widenOpenSesame adds multi-topic AI search for trainingLegora launches legal research foundation for AI lawThrive bolsters NextGen platform with Elastic, Cato
