Introducing Compass: Data-driven decisions right in Slack

It's 9 PM on a Tuesday, and the Slack notification pings. It's the Head of Marketing: "The Q3 campaign dashboard shows a 15% drop in conversions this week. What's going on?" You built the dashboard, you know the data is accurate, but you also know the chart can't answer the why. The real answer lies in a dozen follow-up queries you'll have to run tomorrow, weaving in context that lives outside the BI tool. This gap between the static chart and the dynamic conversation needed to understand it is where self-service analytics breaks down.

Tools that provide static answers to predefined questions don't help people think and reason with data. True data fluency emerges only when people can bring their own context, institutional knowledge, and evolving questions into a dialogue with the data itself and engage with it on a deeper level.

Why Dashboards Don't Help

‍

Think about the analysts who craft dashboards. No chart is the result of a single query. The actual work, exploration, iteration, and dozens of alternative queries tested along the way, is hidden from the end user. That process of exploratory data analysis is where the real insights are first obtained.

Most people instinctively know what they want from the data, but the real barrier is technical skill. Analysts become the gatekeepers of data because they spend time with it and have the expertise to translate thoughts into queries that extract meaning.

Crafting questions, interpreting results, and refining those questions in a loop is what makes data a true two-sided conversation. This highly iterative approach embodies data fluency but it cannot exist in dashboards. Dashboards can only ever tell you the what of data, never the why.

At their core, dashboards are built for technical audiences and live outside the normal workflows of most users. They also lack an interface for non-technical experts to contribute their knowledge back to the system. If you’re in marketing and notice a spike in a graph, you may immediately know that a recent campaign makes that data point irrelevant. But how do you tell that to the dashboard?

This limitation prevents meaningful engagement with data and silos critical context. Even if the dashboard is correct, it doesn’t create a shared picture across the organization. Meeting users where they actually work, in Slack and not a separate reporting tool, allows data to become part of daily conversations. Without that, people stay passive, their thoughts about data detached from their actual workflows. A unique approach to context management is needed to bring the why into the discussion and make data engagement genuinely collaborative.

How AI Conversations Change the Game

This one-sided, frustrating loop is a choice, not a necessity. What if, instead of being a passive consumer of pre-built charts, you could enter into a genuine dialogue with your data? This is where the paradigm shifts, powered by recent advances in AI. Instead of the single prepackaged answer, users can engage in a back-and-forth dialogue with their data. You can ask follow-up questions, refine assumptions, and incorporate your own institutional knowledge directly into the analysis.

By expanding the number of people who can have two-sided conversations with data, more people are encouraged to dive deeper and provide their own understanding of the data to the system. Removing the technical barrier also provides the feedback loop that historically has only been available to analysts. Instead of submitting a ticket and requesting additional information, they can pursue everything themselves. That loop builds understanding over time, which is the essence of data fluency.

‍

“A lot of times these big questions … you spend a week making sure it’s perfect, but really all they wanted was just a directionally right answer.”

— Data Analyst, Design Partner

The partnership between people and machines amplifies understanding by combining the best traits of both. Humans bring domain knowledge, historical context, and the ability to spot nuances that might be lost in raw data. AI systems bring computational scale, memory, and the ability to surface relationships and anomalies across vast amounts of data.

The real opportunity lies in how business context is captured and managed. Context is often fragmented across dashboards, tribal knowledge, and ad-hoc conversations. If it isn't systematized, the same questions get asked repeatedly, and organizations never truly compound their knowledge.

This changes the entire relationship with the data team. Instead of requiring analysts to know everything about the organization, they can focus on the health of the data platform and allow experts from across the organization to mine and provide insights.

The best place to start managing this context is through software engineering practices. Treating metrics, definitions, and assumptions as version-controlled artifacts ensures they are reviewable, testable, and consistently applied across teams.

But we also expect this to evolve. Over time, conversational interfaces will likely become richer repositories of business knowledge themselves. Not only querying data, but also ingesting context as part of the workflow. Each conversation will add to the organization's collective memory, making both humans and machines sharper the next time around.

Compass

‍

We believe that Compass is the natural next step for Dagster. Dagster was built to tame the complexity of modern data applications, pipelines that span countless systems, teams, and moving parts. We've helped customers harness this complexity and build data platforms that do remarkable things.

But while Dagster manages the infrastructure and applications, the end-user experience with data has remained underserved. Dashboards, no matter how polished, will never empower users to keep pace with the complexity of modern organizations. They leave people always trying to interpret someone else's framing of the data.

Compass is an extension of our mission. Just as Dagster gave data teams a framework to manage data complexity on the backend, Compass provides end users with the tools to navigate analytical complexity on the front end. It's not about more dashboards, it's about empowering users to converse with their data, bring their own knowledge to the table, and grow more fluent with every interaction.

‍

“Compass has been a total game changer for me. I save at least half a day a week that I used to spend digging through internal and external data and building charts for ad hoc team requests. It’s automated our pipeline reporting and made data way more accessible, so anyone on the team can dig in, be curious, and get insights. It’s opened the door for more analysis and deeper thinking than we ever had before.”

— RevOps Leader, Design Partner

What's Next for Compass

This is just the beginning. Our vision for Compass is to evolve beyond a query interface into an active repository of organizational knowledge. In the future, Compass will not only answer questions but also proactively learn from every interaction, making your organization's collective data memory sharper and more accessible with each conversation.

Ready to transform how your team engages with data?

• 🚀 Join the Waitlist: Want early access? Sign up to be among the first to bring conversational analytics to your organization.
• 🌐 Explore Compass: Learn how to ask questions in plain language and get instant insights, definitions, and analysis.