RootCause.ai's data ontology connects your fragmented, siloed sources and resolves them into one consistent model: entities, time, and metrics defined once, agreed across every system.
Fragmented data
How do all the datasets calculate churn?
Three different rules, one per source:
churn: No deals closed in 90 days
churn: No orders fulfilled in 60 days
churn: No purchase on any channel in 30 days
Standardise churn to no orders on any channel for 90 days.
churn: No order on any channel in 90 days
Revenue is one number in finance and another in sales ops. The same customer appears four times under four keys, so a headcount of customers is wrong before anyone asks for it.
An agent inherits that disagreement. Every answer it gives rests on whichever definition it happened to read first. The answers stay confident and stop agreeing, and every system you connect widens the gap.
Your data sits in different systems that were never built to work together. A warehouse here, a few databases there, files and exports on the side. RootCause pulls them into one model, so you're working from a single view instead of stitching sources together by hand every time.
Ask three systems what "active customer" means and you get three answers. RootCause picks one.
Every metric gets a single definition, a named owner, and the aliases your team already says out loud. Table relationships are verified against real data instead of inferred from column names. Column meaning is read in context, so a negative value that signals a processing error in one table doesn't get counted as a refund in another.
RootCause matches the identifiers, then checks each match against evidence already on the record: domain, phone, billing address, account history. Matches with enough agreement are merged. The rest go to a person.
Ask a weekly question of hourly and monthly data and they don't line up. RootCause aligns event timestamps, daily snapshots and fiscal periods to the period you asked for, applying the right timezone and fiscal calendar. The alignment is saved to the ontology, so the next question inherits it.
RootCause maps points, postal codes, sites, cities, regions and countries onto one hierarchy. Ask at country level and every coordinate underneath rolls up to it.
An ontology driven agent queries a semantic layer that maintains itself. New columns get read, new sources get resolved into the same entities, and definitions hold without anyone rewriting them. Ask in plain language and get one number back, at the agreed definition, with the rows behind it. Connect through API, SDK, and MCP.
What was churn last quarter, on the company definition?
Databases, warehouses, object storage, SaaS applications and flat files all feed the same ontology. RootCause connects with read-only credentials scoped to what you nominate.
Transactional row stores and read replicas, read with credentials scoped to the tables you nominate.
Modeled tables and raw landing zones alike. An existing dbt project is welcome and never required.
Columnar and delimited files in buckets, including the partitioned exports nothing else reads.
CRM, support, billing and marketing systems, pulled through their own APIs with their own field semantics.
Spreadsheets and extracts that carry real business logic and live on somebody's drive.
Bring one messy source and one you care about. We will build the ontology on the call.