Conversational analytics, built on data you trust
Ask a question in plain English. Get an answer backed by your governed metrics, tested models, and business logic—not a guess.
Conversational AI is only as reliable as the data behind it
Without a governed context layer, AI tools are guessing at your metrics. Here's what that costs—and what changes when you fix it.
Metrics that mean different things in different tools
Your revenue figure in Salesforce, Tableau, and your AI summary are three different numbers. Without a shared definition layer, every tool is running its own logic.
AI outputs that can't be traced or explained
The AI returned a number. You don't know which table it pulled, which logic it applied, or whether that logic is current. When it's wrong, debugging is guesswork.
Data teams stuck fielding the same questions
Business users need answers they can't get without engineering help. Tickets pile up, the backlog grows, and the work that actually moves the business forward keeps getting pushed.
Consistent metric definitions, everywhere AI queries them
dbt Semantic Layer defines your KPIs once—revenue, conversion rate, churn—and delivers that same definition to every AI tool, BI platform, and workflow your team uses.
Every answer traceable back to its source
The dbt MCP server gives AI access to your lineage, model documentation, and metadata—so any answer it surfaces can be traced back to the data and logic behind it.
Business users who can self-serve, without the risk
When AI has governed dbt context underneath it, business teams get accurate, consistent answers on their own—without waiting on a data engineer or second-guessing the result.
Teams are already shipping conversational analytics with dbt
After connecting dbt to Claude via the Semantic Layer and MCP server, Sweetgreen's business teams get consistent, governed answers on their own—without looping in a single data engineer.
By leveraging AI and the dbt Semantic Layer, self-service analysis has become a 30-minute job, compared to the old process of reaching out to the data team and waiting two weeks for the bandwidth to get an answer.
Sankalp Vatsh Analytics Lead @Sweetgreen
Our ambition is to have a few standard reports with core metrics combined with an innovative solution that enables anyone at Norlys to retrieve the insights they need simply by asking a question through a chatbot interface powered by an LLM. dbt MCP server and dbt Semantic Layer enable this.
Søren Persson Director of Data Engineering at Norlys
dbt provides the governed context with metrics, lineage, and tests, and the dbt MCP server makes that context usable by AI systems. Together, they let us design and set up trustworthy conversational analytics quickly for customers—with an audit trail by default.
Jonas Munk Partner at LEAP Consulting
Built for teams making their data AI-ready
If any of these describe your team, dbt is the right foundation.
Evaluating AI-powered analytics tools
You're piloting LLM-powered BI, copilots, or natural language querying — and you don't want your metric definitions locked inside whichever tool wins. dbt gives you one structured context layer that every AI interface reasons over, so switching tools never means rebuilding your governance.
Getting inconsistent AI answers
Snowflake Cortex, Databricks AI, and BigQuery Duet generate inconsistent answers when metric logic isn't defined once and reused. dbt is the structured context layer underneath — governed metrics, tested models, lineage — that turns warehouse-native AI from plausible into trustworthy.
Stuck with governance that doesn't reach AI
Your team already invested in metric definitions — in LookML, a wiki, YAML files, or a homegrown semantic layer — but none of it is accessible to the AI tools you're being asked to deploy. dbt centralizes your context layer in version-controlled code and exposes it through the MCP server, so the governance you've already built powers every AI interface.
See conversational analytics built on your data
Get a walkthrough of how dbt connects your structured context to any AI tool—so business teams get governed, accurate answers, and your data team stops being the bottleneck.
Governed metrics, ready for AI
See how dbt Semantic Layer defines your KPIs once and makes them available to any LLM or BI tool.
Answers your team can actually trust
Learn how AI powered by dbt traces every answer back to lineage, logic, and tested models.
Self-service that sticks
See how teams like Sweetgreen went from 2-week data requests to 30-minute self-service analysis.
Build conversational analytics your business can trust
Connect any LLM to your governed dbt data with the dbt MCP server, available now.