Ask your data. Trust the answer.
Dex is the conversational AI experience built on DataOS. Ask a business question in plain language and get governed, explainable answers grounded in the business context, semantics, and policies of your Data products.
.svg)
Dex is the conversation. DataOS is the foundation.
Connecting AI to your data is easy. Getting answers you can use is not.
Most companies have invested heavily in data platforms, yet business users still wait on dashboards, analysts, and data engineers to answer everyday questions. Plugging a general-purpose LLM into that data gives it access, but access is a long way from understanding.
What Dex gives your organization
Keep asking. Dex keeps the thread.
Real analysis is rarely one question. Dex remembers where the conversation has been, so each follow-up builds on the last answer instead of starting over.
Why did conversion drop last week?
Was it concentrated in certain regions?
How much was related to product availability?
Show the top five regions as a chart.
Behind every answer, Dex shows how it got there: the metric and its definition, the Data product it used, and the filters it applied. If the trusted data can't answer the question, Dex tells you instead of guessing.
From question to governed answer in four steps
Built for everyone who asks questions of data
General assistants are built to always produce an answer, even from unverified sources. Dex only answers from governed Data products and says so when they don't cover the question.
Dex doesn't require your data to live in a single vendor's ecosystem or behind a specific BI report. It works across the platforms you already use.
Use commercial, enterprise, or private LLMs, and switch as your requirements for security, privacy, latency, and cost change.
Frequently Asked
Questions
A connection gives the model access. Dex adds the business meaning: it uses DataOS to supply the relevant data product, its definitions, relationships, and policies, so the model reasons over focused, governed context instead of raw tables.
No. Dex works on top of the platforms you already run, including Snowflake, Databricks, BigQuery, Iceberg, databases, and APIs. It complements dashboards and is most useful when the next question isn't already answered by one.
Governance is enforced in DataOS, not left to the language model. The user's identity and access policies are evaluated during the query, so every answer reflects only what that person is authorized to see.
Yes. Answers include the metric used, its definition, the data product, the filters applied, and the underlying sources.
The one that fits your AI strategy. Dex supports commercial, enterprise, and private models, so you can balance capability, privacy, latency, and cost.
Dex sends the model only the context that's relevant to each question, which keeps prompts smaller. It also tracks models, messages, tokens, and spend, giving you the basis to see consumption by user, team, Data product, or use case.
A data product narrows the scope from everything you own to a governed set of data designed for a business purpose. That gives Dex clear definitions and controls to work with, rather than asking an LLM to infer meaning from raw schemas.