Introducing dbt Copilot: The future of AI-accelerated analytics

Today at Coalesce 2024, we proudly unveiled dbt Copilot—the AI engine embedded within dbt Cloud to accelerate your analytics workflows. dbt Copilot automatically generates documentation, semantic models, and data tests, while also offering powerful natural language chat, enabling any stakeholder to easily interact with your data. And this is just the beginning. Starting today, dbt Copilot is available in beta.

The future of AI-accelerated analytics

As a data practitioner, you've likely noticed the growing trend—analytics teams are increasingly adopting software development best practices to ensure their data is reliable, scalable, governed, and future-proof. We call this the analytics development lifecycle (ADLC), a scalable process designed to streamline and optimize data operations. By incorporating these best practices into your analytics workflows, you can deliver high-quality, trusted data at speed and scale.

But, you shouldn’t have to do it all on your own.

We believe practitioners should be focusing on higher-impact projects rather than getting bogged down by repetitive tasks like debugging, figuring out the right syntax for SQL functions, writing tests, and answering the same data questions repeatedly. While these tasks are necessary, they shouldn’t dominate your time. That’s where AI can act as a powerful assistant, automating routine processes and freeing you to tackle more complex, strategic challenges.

That’s why we are excited to introduce dbt Copilot, the AI engine embedded within dbt Cloud designed to accelerate your analytics workflows.

Now available in beta, dbt Copilot seamlessly integrates AI-powered assistance throughout your dbt Cloud experience, empowering you to ship data products faster, and deliver higher data quality with confidence. It takes care of the tedious tasks—like drafting documentation, building semantic models, authoring data tests, and answering well-understood data questions—within a strong best-practice framework. Most importantly, dbt Copilot keeps you in control, with a human always in the loop. The AI isn't replacing you; it’s enhancing your workflow, allowing you to use your expertise to oversee and guide the process, ensuring quality and context are preserved.

"My CFO doesn’t want to see a dashboard—she wants direct answers, like our ARR for the quarter. With dbt Copilot, she can instantly query well-defined metrics using conversational language so she can get the answers she needs instantly and my teams can get out of the business of building dashboards. The future of analytics is AI-powered insights and governed metrics via Semantic Layer, not dashboards. As these tools evolve, organizations that embrace them will lead the way."

Josh CarlsonDirector of Analytics

Accelerating DataOps across the ADLC with dbt Copilot

Our goal with dbt Copilot is to deliver AI to make data practitioners more productive by accelerating every stage of the ADLC. Here’s how dbt Copilot is helping today and what’s coming soon:

Develop: Automatically generate documentation and code

You’re likely already managing documentation and building semantic models, or planning to implement these analytics best practices. But tasks like documenting 100 columns or defining dozens of metrics manually can be repetitive and time-consuming.

Auto-generate documentation

Today, dbt Copilot leverages generative AI to automatically draft your documentation, speeding up the process and allowing you to quickly review and refine the content to meet your standards. This enables you uphold best practices with just a click.

Auto-generate semantic models

When it comes to semantic models, defining key business metrics with precision is essential for your business, but it doesn’t need to be a long, manual task. dbt Copilot builds an initial draft of your semantic models and the metrics that power it, to help you get up and running with the dbt Semantic Layer faster. With dbt Copilot handing the foundational code, you can focus on refining and aligning your team’s efforts.

Test: Auto-generate tests

We know that many dbt models have low test coverage, and a big reason for that is the tedious nature of writing tests. Defining tests can feel like manual labor—identifying primary and foreign keys, scaffolding YAML code, and profiling datasets can be time-consuming. But without proper testing, you risk data quality issues that can erode trust.

Now, dbt Copilot automatically generates a baseline set of data tests, helping you catch potential errors in real-time across your entire DAG. This way, you can ensure data quality without getting bogged down by the manual work, allowing you to focus on building while staying ahead of issues before your stakeholders even notice them.

Looking ahead, dbt Copilot will soon help you create unit tests in the dbt Cloud IDE. Using generative AI, dbt Copilot will scaffold out mock input data and expected output data that will stress-tests tricky transformation logic like date math, regular expressions, and long case-when statements. These unit tests will ensure your logic is valid, cutting down on troubleshooting time and ensuring the correctness of your transformations.

Analyze: Chat with your data

Answering data questions from stakeholders often requires frequent context-switching and tedious work. Centralizing business metrics in the dbt Semantic Layer has simplified this by allowing users to rigorously define key metrics like ARR, churn rate, and WAUs, making it easier to provide stakeholders with self-service access to the data they need.

Now, we’re taking it a step further with dbt Copilot. In addition to querying governed metrics from a BI tool, stakeholder can now ask their questions in conversational language, turning the process into a seamless "chat with your data" experience. This not only delivers instant insights to your stakeholders but also frees up your time to focus on higher-impact work.

Looking ahead, you can use dbt Copilot to discover trusted data sets more efficiently with AI-powered semantic search in dbt Cloud.

The future of analytics starts now with AI-driven workflows

By integrating AI into every stage of the ADLC, dbt Copilot automates the tedious tasks, freeing you to focus on strategic, high-impact work.

Currently in beta, dbt Copilot offers key features including:

• Auto-generated documentation to speed up documentation creation and review.
• Auto-generated semantic models and metrics for faster adoption of the Semantic Layer.
• Auto-generated data tests to ensure data quality in real-time.
• “Chat with your data” for natural language querying of well-defined metrics to provide instant insights to any downstream data stakeholder.

In the future, dbt Copilot will include even more powerful tools, such as natural language SQL generation to accelerate development, AI-powered unit tests, and much more. We're excited about how these features will help you achieve more with your data, and we're committed to expanding AI-powered support across all stages of the ADLC. If you’re interested in a participating in our beta please sign up here.

To learn more, be sure to join our upcoming virtual event to dive deeper into the new dbt features announced at Coalesce 2024, including dbt Copilot. We'll have live demos and experts ready to help you make the most of these new features, so bring your questions and join the conversation. Register here now to watch live or receive the recording.