Insightsoftware unifies semantic layer, governance to aid AI
With many enterprises struggling to develop AI tools, the vendor's Simba Intelligence feature set is designed to feed agents and other applications trustworthy, transparent data.
Insightsoftware has launched Simba Intelligence, a feature set designed to enable customers to develop AI applications using trusted data that includes a semantic layer unified with automatically applied governance and security policies.
Semantic models are sets of common definitions of data characteristics. When applied across an enterprise's data, they make data consistent so it can be searched, discovered and integrated to inform models and applications, including AI chatbots and agents.
Semantic modeling dates back to the 1970s, and it has been used as part of the analytics process for more than two decades. However, its popularity has markedly increased over the past couple of years as more enterprises are investing in AI development.
AI tools require enormous volumes of high-quality data to be accurate. In addition, they need fresh, real-time data to be current. When business decisions and processes were informed by rear-facing reports and dashboards, humans could handle data preparation, and semantic models were a luxury rather than a requirement.
Now, however, AI is making semantic modeling more necessary, according to William McKnight, president of McKnight Consulting. He noted that large language models need governed, consistent data when querying massive data environments or else they will hallucinate.
"Without that control layer, scale will amplify any inconsistency," McKnight said.
Simba Intelligence, made generally available on March 3, features a semantic layer to make data consistent and discoverable. Meanwhile, by pairing semantic modeling with governance and security policies and connecting it directly to an enterprise's proprietary data, Insightsoftware's new feature set makes AI outputs transparent and trustworthy.
As a result, Simba Intelligence is a valuable addition for Insightsoftware customers, according to McKnight.
"Simba Intelligence is a solid step for Insightsoftware towards relevancy in agentic AI," he said. "They are making it possible to conduct agentic AI on the imperfect data that every company has."
Based in Raleigh, N.C., Insightsoftware is a former ERP specialist that expanded into data management and analytics a series of acquisitions, beginning with its April 2021 purchase of embedded BI specialist Logi Analytics. Beyond Insightsoftware, DBT Labs, Cube, Google's Looker and ThoughtSpot are among the vendors that include semantic modeling capabilities as part of their offerings.
In addition, with each vendor's semantic modeling capabilities differing from one another, a group of vendors including Snowflake and Salesforce in September 2025 formed the Open Semantic Interchange to develop an open-source standard for semantic data modeling.
Enabling AI
Although enterprises have been increasingly pouring money into AI development since OpenAI's November 2022 launch of ChatGPT marked significant improvement in generative AI technology, most initiatives don't make it past the pilot stage.
Simba Intelligence is a solid step for Insightsoftware towards relevancy in agentic AI. They are making it possible to conduct agentic AI on the imperfect data that every company has.
William McKnightPresident, McKnight Consulting
While there are myriad reasons AI projects stall before making it into production, primary problems enterprises face include disorganized data that makes relevant data difficult to discover and poor data quality that make AI outputs unreliable.
Numerous data management vendors have responded since the start of 2026 by launching new capabilities specifically aimed at helping customers more successfully develop AI tools.
For example, Databricks released Instructed Retriever, an alternative to standard retrieval-augmented generation that targets the data discovery process. MongoDB launched a series of vector embedding and ranking models that likewise are designed to improve data retrieval to make AI applications more accurate and trustworthy. And Domo, Teradata and Graphwise have all similarly introduced new features designed to help users not only begin AI development projects but put them into production.
Now, Insightsoftware is doing the same with the launch of Simba Intelligence in response to user feedback, according to Matt Belkin, the vendor's president of Data + Analytics business unit.
"Most enterprises don't trust their AI, and they've been telling us why," he said. "AI pilots look great in demos and fall apart in production because models hallucinate against data they don't understand. Without business context, AI is just guessing confidently. That's exactly what the semantic layer solves. It gives AI the meaning behind the data, not just the data itself."
Simba Intelligence is, in effect, a semantic modeling and governance layer that sits between data sources and AI tools.
By automatically applying an enterprise's semantic definitions, governance and security policies and providing audit trails, the feature set makes sure AI outputs are consistent and traceable, and that they are in alignment with organizational policies. In addition, positioned between data sources and AI tools, Simba Intelligence can query data wherever it is stored -- including cloud, on-premises and hybrid environments -- without forcing customers to move data from one system to another.
Matt Aslett, an analyst at ISG Software Research, noted that while semantic modeling is valuable, it has historically been viewed as a niche activity. Even last year, just 9% of participants in ISG's Market Lens Data and AI survey named semantic modeling as one of their five largest-funded data activities.
Now, however, trends including natural language search, which often use semantic-based knowledge graphs to improve retrieval accuracy, and agentic AI, which needs the business context that semantic modeling can provide, are increasing popularity of semantic layers.
"ISG asserts that through 2027, analytics and data providers will prioritize the development of semantic modeling capabilities to facilitate the execution of autonomous AI agents," Aslett said.
Consequently, Simba Intelligence is a valuable new feature for Insightsoftware users, he continued.
"The addition of semantic layer to Simba Intelligence will better enable customers of Insightsoftware’s Data + Analytics business unit to provide AI initiatives with trusted, business-ready data," Aslett said.
Looking ahead
As Insightsoftware's plans its data management and analytics roadmap, enabling customers to develop and deploy AI tools in on-premises environments is one priority, according to Belkin.
Most AI development is done in the cloud given its scalability, flexibility and compute power. However, many enterprises in highly regulated industries such as defense, financial services and healthcare keep much of their sensitive information on premises due to security concerns related to the cloud.
"That market is underserved," Belkin said. "We are investing in giving organizations full control over data residency without giving up AI capability."
In addition, Insightsoftware plans to add more feature sets similar to Simba Intelligence that are designed to better enable customers to move AI projects into production, he continued.
Aslett noted that InsightSoftware's Data + Analytics unit has, to date, limited its focus to core connectivity through master data management and insight generation through data visualization. To better serve the needs of current users and perhaps even attract new ones, he suggested that it could expand into other areas of data management.
"Opportunities for expansion include data governance, data quality and data pipeline development and orchestration, as well as data observability," Aslett said.
McKnight, meanwhile, recommended that Insightsoftware advance the embedded analytics capabilities it provides under the Logi brand to include embedded AI agents. Most embedded analytics capabilities act in response to user queries or instructions. Embedded agents would alter the paradigm, enabling embedded tools to act on data without being prompted first.
"Someone could lead this field by offering task-specific AI agents that act on data," McKnight said. "Another winner would be to create the ability to take a complex Excel model and instantly turn it into a governed, multi-user web application with full write-back and audit trails."
Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.
