An example: the number 4,206 answers “How many active customers did we have last month?”. Depositori knows that an active customer is one with an order in the last 90 days, that the number lives in Snowflake, and how it was derived.
Every data point has a story.
The semantic layer for the stack you already run. Ask in plain language; the query runs on the source, and only the answer comes back.
Request early access- Power BI
- dbt
- Snowflake
- Databricks
- PostgreSQL
- SharePoint
- Confluence
- SAP
- Google Analytics
- Excel
- Salesforce
- Microsoft Fabric
- Tableau
- Cube
- Omni
- Sigma
- ThoughtSpot
- Hex
- Holistics
- GoodData
- Honeydew
- WisdomAI
- NVIDIA Auto Ontology
- Palantir
- RelationalAI
- Apache Polaris
- OrionBelt
Why Depositori
One glossary, and a map of where each term lives.
Depositori reads every definition, measure and lineage from Power BI, dbt, Snowflake, PostgreSQL, SharePoint and the rest, and turns them into one glossary. No new database, no migration.
The glossary says what each term means. The semantic layer records where each term lives and how the terms relate, so any tool or agent can use them.
active customer
ApprovedIn words
A customer who placed an order in the last 90 days.
active customer
ApprovedIn words
A customer who placed an order in the last 90 days.
As a query
select count(distinct customer_id)from orderswhere order_date >= current_date - 90Uses it
- Power BI · Customer dashboard4,206
- dbt · active_customers4,206
- Agent · “How many active customers?”4,206
Owners
- MAMarketing analystProposed 60 days
- FMFinance managerReviewing
One term, said every day.
Marketing, sales and finance all talk about active customers. The glossary writes down what it means, in words anyone can read.
The same definition, as a query.
Under the words sits the query that counts it. Depositori keeps the two together, so what people say and what the numbers show can't drift apart.
Everything that uses it.
The customer dashboard in Power BI, the dbt model and every agent's answer read this one definition. Right now they all say 4,206.
A change waits for its owners.
Marketing proposes 60 days instead of 90. Finance owns the term too, and until the finance manager approves, nothing changes anywhere.
Change it once.
The finance manager approves. The words, the query, the dashboard, the model and the agent all move to 60 days together.
Or Finance says no.
Had Finance rejected the change, active customer would stay at 90 days in every report. Marketing still gets its number: the 60-day count is added to the glossary as a metric of its own, recently active customer.
How it works
Plug in. Know where it lives. Ask. Match.
Plug in
Point Depositori at your stack and it reads the definitions, measures and lineage in each system. Out comes one glossary of all your data, built in one go.
Know where it lives
The glossary records every place a term lives: the Power BI measure, the dbt metric and the table they are built on, with the lineage between them. So Depositori knows which sources hold an answer before anyone asks.
Ask
Ask in plain language, in Depositori's own agent or one you already use. When a term lives in more than one place, Depositori runs the question on each source, in its own language, and puts the answers side by side. Only the results come back.
Net revenue
- Reviewed
Glossary
Sales − Returns − Discounts
- Differs
Power BI
Sales − Returns
- Matches
dbt
Sales − Returns − Discounts
Pick the definition that holds.
Match
Where the answers differ, Depositori shows each source's definition next to the glossary's. Someone on your team decides which one holds, and every answer after that uses it.
Migrate with AI, check every number.
Moving from Power BI to dbt, Databricks or Snowflake? Depositori translates each model, lets AI draft the measures its rules can't translate, and checks that the new numbers match the old ones. Every translation lists what didn't carry over.
Any agent can plug in.
Claude, ChatGPT, Copilot or your own agent connects over MCP or the API, each with its own key. Agents can ask for reviewed definitions only, so they answer with the ones your team agreed on.
Your data stays where it is.
Depositori reads definitions and metadata, never rows, and writes nothing back to your systems. Queries run with your own credentials. The AI sees definitions and aggregated results, and columns with personal data are refused.
The name
Deposi-: From depository, where you keep what is valuable.
-tori: Sranantongo for story.
Tell us about your stack.
We read every message and reply personally.
Where we are
Our office is on the Herengracht, on Amsterdam's canal ring.