Process documents
Bring in policies, procedures and data dictionaries. Kenseme reads them and proposes the terms they depend on.
Platform
Kenseme captures what your business means, models it as an ontology, binds it to your data, governs it, and puts it to work in the data agent or the agent you already run. Five areas, one model running through all of them.
01 · Document meaning
Most of what your business means is already written down, scattered across policies, procedures, data dictionaries and glossaries. Kenseme reads those sources and turns them into defined, sourced terms.
Bring in policies, procedures and data dictionaries. Kenseme reads them and proposes the terms they depend on.
Each proposed term arrives with a definition and the source it came from, ready for your experts to review.
Anchor your vocabulary to the open standards for your industry instead of starting from scratch.
02 · Build the model
An ontology is a formal description of your business: the concepts, how they relate, and the rules between them. Kenseme builds one from your glossary, your documents and the open standards for your industry.
Draft an ontology from your glossary, documents and open industry standards in one pass.
Describe the change you want in plain language, add a concept, split one in two, explain a relationship, and Kenseme proposes the edit.
The diagram view shows every concept and relationship, so reviewers can spot what's missing.
Export all of it whenever you like.
Start from proven structures for your industry rather than a blank canvas.
Every change lands in a version history you can review.
03 · Bind to your data and tools
Bind the ontology to your warehouse or lakehouse and to the tools that act on your systems, and design the data that doesn't exist yet. Every run shows a preview first, so nothing changes without someone seeing it.
Map concepts and properties to tables and columns in a database connection or a Fabric Lakehouse.
Map the actions in the model to the tools that carry them out in your systems, so an agent knows which one to call.
Publish the bound ontology to Fabric, then push or pull changes, with a preview before every run.
Create, version and deploy data models, including AI-generated star schemas.
SQL with procedures for SQL Server, Azure SQL, Fabric Warehouse, Postgres, MySQL, Redshift and Snowflake. PySpark notebooks for Databricks and Fabric Lakehouse.
Generate the pipelines that bring data into the model, in your platform's own dialect.
Define the transformations that move data from one model to the next.
Binding works with database connections and Fabric Lakehouse. Code generation covers all eight platforms listed above.
04 · Govern
Definitions, owners and versions live with the model itself, not in a separate spreadsheet. Govern your data assets from it, then publish to the governance stores you run, such as Unity Catalog.
Models, assets and terms in one governed catalog, published to more than one governance store, such as Unity Catalog.
One place for every governed term, its definition and where it came from.
Version history on every model, and custom attributes for owner, steward, review cycle or any field you use.
05 · Agent action
Data agents answer questions, and other agents act on your systems through tools. Give both the same governed meaning, whether they run in Kenseme or on your cloud data platform: Microsoft IQ, Databricks Genie, Snowflake Cortex, Amazon Quick Suite, Gemini Enterprise or Oracle Select AI.
Ask a business question in plain language. The data agent runs it against your bound data.
Every answer comes back with a grid, a chart and the exact SQL, so anyone can check it.
Save the questions you ask often and come back to them.
Agents act on your systems through the tools bound to the model, with each action defined in business terms.
Deploy the model as an ontology for the data agents and operations agents you build in Fabric.
Expose the ontology over MCP so other agents and tools can read it.
Together we pick one business area, the one where a wrong answer costs the most. Two weeks later you can ask a real question and check the answer.