| Kyvos | |
|---|---|
 | |
| Type | Private |
| Founded | June 30, 2015 |
| Founder | Praveen Kankariya |
| Products | Semantic layer Stable release: Kyvos 2026.6 License: Proprietary |
| Website | kyvosinsights |
Kyvos is a semantic layer for AI and BI, developed by an American privately held company named Kyvos Insights and launched officially in June 2015. The company, headquartered in San Francisco, California, was founded by Praveen Kankariya.
Kyvos semantic layer sits between enterprise data platforms and the tools that query them, including AI agents, analytics applications and business intelligence tools. Kyvos centrally defines and governs business definitions, dimensions, attributes, hierarchies, metrics, and business logic.
For AI, Kyvos provides context by combining semantics with unstructured knowledge and business relationships.
Kyvos helps build a business knowledge graph that maps KPIs, entities, concepts, and knowledge, with the rules, weights, and causal paths that connect them.
Kyvos semantic models define dimensions and attributes, metrics and calculations, and business names and descriptions for data.
Kyvos converts unstructured knowledge, such as policies, procedures and product documentation, into vector embeddings. For each query, it retrieves the relevant knowledge from these embeddings using retrieval-augmented generation (RAG). The retrieved knowledge becomes part of the business context that Kyvos assembles for the AI agent or application.
Kyvos combines semantics, unstructured knowledge and business relationships into context for AI. Kyvos AI Space is the workspace to configure this context for a specific AI use case, along with the analytics tools that AI agents use to reason and act.
Dialogs is Kyvos’ business analytics agent, launched in May 2025. It interprets business questions, resolves them against the context in an AI Space and runs the required queries. Dialogs explains the drivers behind a result, supports multi-turn analysis and shows how each answer was derived. Other AI agents can call Dialogs through MCP, and it is available in Microsoft Teams and Slack.
Kyvos semantic layer maintains data access rules alongside business metrics and calculation logic. Its authentication and authorization component, AuthK, supports single sign-on and the assignment of roles to users and groups to control access to data, reports and features. It also includes data masking, encryption of data at rest and in transit, and auditing and logging of activity.
Kyvos uses distributed query execution, smart aggregation, caching, and elastic compute to process queries on large datasets.
In August 2026, Kyvos announced compatibility with Apache Ossie, an open-source project for standardizing semantic model exchange across analytics, AI and BI platforms.
Kyvos provides a model context protocol (MCP) server for connecting AI agents and applications to Kyvos. It also supports integration with the LangChain framework.
Kyvos connects to Power BI, Tableau, Looker, Strategy (formerly MicroStrategy) and Excel through dedicated connectors. BI tools can also connect to Kyvos using SQL, MDX and DAX.
Kyvos also provides REST and Java APIs for programmatic access from other applications.
Kyvos supports Amazon Web Services (AWS), Cloudera, Google Cloud, Microsoft Azure, Redshift, Snowflake, Databricks, and Google BigQuery.