Described, not guessed

An AI that actually knows your business.

Kenseme turns your business knowledge into a model every agent can work from: your terms, processes, rules, and what your people know but never wrote down. Your experts govern it. Agents use it to find the data behind each term and run your processes through your MCP tools, alongside your team.

2.2×

more answers rated excellent, with the same model, grounded in an ontology versus not

21%

of tasks solved against real enterprise schemas, by models that score 91% on a simple one

30%

fewer tool calls to reach the answer, which is what guessing costs

Microsoft-controlled A/B, 4,000 responses across 400 questions, Fabric Updates Blog, June 2026 · Spider 2.0, ICLR 2025

Book a demo See the platform

Already using Kenseme? Log in.

The problem

Your agents have your tables and your documents. The agents don't have your unique business knowledge.

The facts are in your tables. The meaning is buried in your processes, SOPs, meeting minutes and glossaries, and in your people's heads. Having the documents isn't the same as knowing what they say, and reading the data alone can't recover it. A schema can show an agent that a customer table exists; it can't tell the agent that an active customer excludes trials, or that a return is negative. Without that, agents answer anyway, confidently. We get that knowledge out of your documents and your people's heads and into a model of your business, bound to your data.

The rules live far from the data

Exclusions and edge cases sit in SOPs, spreadsheets and memory, not next to the data.

Terms mean different things in different rooms

Finance, sales and operations each define a customer differently. An agent can't tell which you meant.

Schemas don't explain themselves

Column names were picked for a load job, not for today's question.

What Kenseme builds

A model of your business that your AI can act on

Kenseme builds a model of your business from what you already have: business processes, SOPs, meeting minutes, glossaries, data dictionaries and the people who know the business. It is grounded in open standards, bound to your live data & tools, and governed alongside it. Five steps take you from your documents to answers you can check.

01 · Document meaning

Start from what you already know, written down or not

Bring in your documents, from policies and SOPs to business processes and meeting minutes, plus glossaries. For what lives only in people's heads, our consulting team runs structured working sessions with your experts, such as a weekly order review, and transcribes them, so that knowledge is captured quickly. From those, Kenseme drafts process models, data dictionaries and business rules, lined up with the open standards for your industry. That is what your model is built from, instead of a blank page.

Transcripts of structured sessions our consultants run with your team, such as a weekly order review
Process models, data dictionaries and business rules drafted from your documents and sessions
Open industry standards to anchor against
Custom attributes for whatever your teams need to track

How document meaning works

Meaning · Retail · illustrative

Upload documents SOPs, policies, minutes, process notes, glossaries and session transcripts

Uploaded · 5 documents

MinutesMonthly operations meetingGlossary, rule
SOPHandle a high risk orderProcess, rules
ProcessProcess a customer orderProcess
SessionWeekly order review, facilitated and transcribedProcess, rules
PolicyFraud prevention policyGlossary, rules

Order risk classificationCertified

Classifies each order by fraud risk. High risk orders go to the risk desk before fulfillment.

#	Order amountCustomer order · decimal	New customerCustomer · boolean	Risk levelstring	Annotation
1	10,000 or more	true	"High"	Large order, new customer
2	10,000 or more	false	"Medium"	Large order, known customer
3	500 to 9,999	true	"Medium"	Mid-range order, new customer
4	500 to 9,999	false	"Low"	Mid-range order, known customer
5	Under 500	-	"Low"	Small order

Decision tableSME: Head of Risk Management

02 · Build the model

Describe how the business actually fits together

The meaning you captured becomes an ontology: concepts, the relationships between them, and the rules that govern them, in a model your people can read and your agents can reason over. AI drafts it from your documents, your experts refine it in plain language, and the diagram shows the whole thing at once.

Drafted by AI from your documents and glossaries
Refined in conversation, in plain language
Yours to export in full, whenever you like

How modeling works

Model · Retail · illustrative

RetailRule: every customer order comprises at least one order line

03 · Bind to your data and tools

Connect each concept to its data, and each action to its tool

Bind the model to your warehouse or lakehouse, column by column, with a preview before every run. Where the data you need doesn't exist yet, Kenseme designs it: schemas, star schemas, pipelines and code for your platform.

Bind to a database connection or a Fabric Lakehouse
Bind the actions in the model to the tools that carry them out in your systems
SQL and PySpark generated for eight platforms, including Fabric, Snowflake and Databricks
Transformation pipelines between models, previewed before every run

How binding works

Bindings · illustrative

Property	Column	Status
Customer.id	`dim_customer.customer_id`	Bound
Customer.isActive	`dim_customer.active_flag`	Bound
Order.date	`fact_sales.order_date`	Bound
Return.amount	`fact_returns.return_amt`	Review

Preview: 4 properties, 3 tables, 0 conflictsReady to run

04 · Govern

One governed catalog, built from work already under way

The catalog fills itself from the modeling, binding and deployment your teams already do. Your stewards add what no scan can find: owners, review cycles and the decisions behind each definition. Use it to govern your data assets, then publish it to the governance stores you run, such as Unity Catalog.

Enterprise data catalog. Models, assets and terms in one governed catalog, published to more than one governance store
Business glossary. Terms drafted from your documents, with owners and stewards
Metadata management. Version history and custom attributes on every model

How governance works

Term · illustrative

Active customer A customer with a completed purchase in the last 12 months. Excludes trial and test accounts.

Owner: Sales operations
Steward: Data governance
Review cycle: Quarterly
Version: 4 (3 earlier versions)
Source: Customer policy, section 2

Unity Catalog and other governance storesPublish

05 · Agent action

Querying your data, or acting on your systems through tools

Data agents answer questions. Other agents act on your systems through tools, using the actions bound to the model. Both are only as good as the meaning they start from, and Kenseme gives them that meaning: governed definitions, rules in business terms, and the columns that hold each one. Run them in Kenseme, or on your cloud data platform: Microsoft IQ, Databricks Genie, Snowflake Cortex, Amazon Quick Suite, Gemini Enterprise or Oracle Select AI.

Data agent. A grid, a chart and the exact SQL behind every answer, with saved reports
Agents that act. They act on your systems through tools, with every action defined in the model and bound to the tool that carries it out
On your platform. Microsoft IQ, Databricks Genie, Snowflake Cortex, Amazon Quick Suite, Gemini Enterprise or Oracle Select AI, all working from the same model
MCP access. Any other agent or tool can read the model

How the data agent works

Agents · illustrative

Which stores had the most returns last quarter?

Downtown$412k

Riverside$355k

Airport$298k

Westgate$241k

SELECT s.store_name, -SUM(r.return_amt) AS returns FROM fact_returns r JOIN dim_store s ON s.store_key = r.store_key WHERE r.return_date >= '2026-04-01' AND r.return_date < '2026-07-01' GROUP BY s.store_name ORDER BY returns DESC;

Keep store returns within policy Watch weekly returns by store. When a store crosses the limit, act through your tools: open a review, tell the store manager, and ask store operations before pausing a promotion.

Watches: Return.amount by Store
Bound to: fact_returns.return_amt
Rule: Weekly returns cross above $50k
Tools: create_case notify pause_promotion

Downtown crossed the limit: $61k this week

create_caseReview return reasons, DowntownDone
notifyStore manager, with the case linkSent
pause_promotionDowntown clearance eventAwaiting approval

Concepts, rules and tools from the modelFabric IQ

Architecture

Where Kenseme sits in your stack

Kenseme sits between your sources of business knowledge and the agents that act on it. Data moves both ways between Kenseme and each data platform you run.

01Source Material

Processes & Rules

SOPs, decision tables, glossaries, business rules

Domain Expertise

SME interviews, tribal knowledge, judgment calls

Data Schemas

ERDs, DDLs, TMDL, data contracts

Industry Standards

The latest standards for your industry

Kenseme on your data platform

02Kenseme

An AI that actually knows your business

02.1Business Meaning

Business Context

Data Context (Ontology)

Governance

Industry Frameworks

02.2Built-in Agents

Ontology & Binding

Ontology Agents

Reasoning Agents

Ingestion Agents

Industry Agents

Data Binding Agents

Briefing & QA Agents

Data Models & Transforms

Data Modeling Agents

Data Pipelines Agents

SQL Agents

Governance & Action

Governance Agents

Data Agents

Action Agents

02.3MCP Server

Exposes all context, metadata, and agent functions via Model Context Protocol

Portable by design Everything you build in Kenseme is portable.

03Data Platforms Cloud data platforms & databases

Microsoft Fabric Azure SQL SQL Server Azure Synapse Databricks Snowflake Amazon Redshift Amazon Aurora Amazon RDS Amazon Athena Oracle Database Oracle Autonomous AI Database MySQL HeatWave PostgreSQL MySQL

04Consumers

Platform Agents

Microsoft IQ (Fabric IQ, Work IQ, Foundry IQ), Databricks Genie, Snowflake Cortex, Amazon Quick Suite, Gemini Enterprise, Oracle Select AI

Kenseme Agents

Kenseme's built-in agents

AI Assistants

Copilot, Claude, ChatGPT

Open by design

Yours to keep, on whatever platform you choose next

Your model is yours, and you can export all of it. Generated code targets the platform you run, in its own dialect. Nothing about your business gets locked inside a tool.

Industry standards

Open standards

Connects to

Microsoft FabricUnity CatalogMCP

Generates code for

SQL ServerAzure SQLFabric WarehouseFabric LakehousePostgresMySQLRedshiftSnowflakeDatabricks

See it on your own data

Two weeks. One business area. Your data.

Start with a demo on sample data, or go straight to a proof of concept on your own data. Together we pick one business area, the one where a wrong answer costs the most, and prove it out end to end. The proof of concept costs a fixed fee, and if you continue within 90 days, the full fee counts toward your first year.

Start the conversation

Pick the business area

We agree on one business area and the questions that matter in it, and gather the documents, glossaries and people who know it best.

Describe and bind

We draft the model from your sources, refine it with your experts, and bind it to your live data.

Ask real questions

Your team asks the questions they actually need answered, in the data agent or the agent you already run, and checks every answer against the SQL.

What we ask of you

One business area, and a sponsor who cares about getting it right
The documents that describe it: policies, glossaries, data dictionaries
Read access to the data that holds it, in a database or a Fabric Lakehouse
A few hours from the people who know the business area best

What you leave with

A model of the business area that you own and can export in full
The model, bound to your live data
A data agent answering real questions, with the SQL behind every answer
A clear view of what the next business area would take

Why one business area?

Because one business area done properly tells you more than ten done roughly. It shows what the model looks like, how your data binds to it, and whether the answers hold up, on a problem your team already cares about.

Which databases do you support?

Kenseme works with the cloud data platforms and databases you already run: Microsoft Fabric, Azure SQL, SQL Server, Azure Synapse, Databricks, Snowflake, Amazon Redshift, Aurora, RDS and Athena, Oracle Database and Oracle Autonomous AI Database, MySQL HeatWave, PostgreSQL and MySQL. Generated code targets SQL Server, Azure SQL, Fabric, PostgreSQL, MySQL, Redshift, Snowflake and Databricks, each in its own dialect.

Where does Kenseme run?

In the Kenseme cloud. Your data stays where it is; queries pass through. If you need Kenseme in your own tenant, talk to us about Enterprise.

What happens after two weeks?

You decide. Extend the model to the next business area, or take it with you. It exports in full, and it is yours either way.

Start the conversation

See what this looks like on your own data

Interested in learning more, or ready to talk about a two-week proof of concept? Leave us your details and we'll come back within one business day. The first step is a short conversation, not a commitment, usually about which business area would prove this out fastest for you.

Prefer email? hello@kenseme.com reaches the same place.