You shouldn’t have to wait days for answers your team needs right now.
And yet, that’s still the norm. Even with all the advances in cloud data warehouses, BI tools, and AI platforms, getting from question to answer still involves friction.
Maybe someone requests a metric from the data team. Maybe they dig through a half-remembered dashboard. Maybe they give up and go with a gut guess. These micro-delays add up, and they’re a major reason why many data investments fall short of their potential.
At Distillery, we kept seeing the same challenge play out: clients had modern data stacks, but no smooth path to access. So we decided to prototype a solution.
Introducing DistillGenie
DistillGenie is a Slack-integrated, AI-powered assistant that connects to governed enterprise data through a semantic layer. It lets users ask business questions in plain English and get back trusted, visual answers instantly.
There’s no SQL. No need to learn a new tool. No pinging your data team for a quick answer.
Under the hood, it uses Databricks Genie as the data engine and AtScale to define and govern business metrics. AtScale’s semantic layer ensures that the numbers are consistent, explainable, and secure, so marketing, finance, and product teams all get the same answer to the same question.
This particular version was built for Slack and Databricks, but the architecture is intentionally flexible. It can be replicated across other platforms like Snowflake, BigQuery, or Redshift, and integrated into Teams, Google Chat, or custom web apps, wherever your team already works.
And it’s not just flexible in where it runs, it’s flexible in how answers are delivered. You can customize the response to show inline tables directly in Slack, link out to a dedicated UI with dynamic charts, include feedback buttons to improve the model, or even expose the exact SQL query used to generate the answer. It’s designed to fit into your workflows, governance preferences, and UX needs.
The Real Problem: It’s Not the Data. It’s the Access.
The more we talked to teams, the clearer it became: access – not storage, not processing – was the real bottleneck.
Even with modern tools in place, teams struggled to get insights into the hands of decision-makers in time. Dashboards were often out of date or difficult to navigate. Data requests clogged up backlog queues. And business users lacked confidence in self-serve tools because the definitions weren’t clear.
That last mile from “I have a question” to “I have the answer” was broken.
DistillGenie was designed to bridge that gap. Not by replacing existing data tools, but by making them more accessible to the people who need them most.
What We Learned Building It
This wasn’t about slapping a chatbot on top of a warehouse. It took careful architecture to make sure the experience was both simple for the user and secure for the organization.
Some key takeaways from building DistillGenie:
- Language parsing is only half the battle. Understanding the question is important, but mapping it to trusted metrics is what makes the answer usable.
- The semantic layer is essential. Without AtScale, we’d risk creating conflicting answers depending on who asked. With it, we define KPIs once and reuse them everywhere.
- Familiar interfaces matter. Putting the assistant in Slack meant zero learning curve, and it worked where people already communicate.
- Scalability depends on modularity. We designed the components (NLQ, data pipeline, governance layer, and UI) so they can be swapped or extended for other environments.
Our team at Distillery handled everything from natural language parsing to Slack interface design, metric mapping, and orchestration across Databricks and AtScale. We didn’t just assemble the pieces. We engineered a system that’s secure, intuitive, and scalable from the ground up.
Why It Matters
For many organizations, the frustration isn’t a lack of data. It’s the inability to get insights into the hands of non-technical teams fast enough. Projects stall. Opportunities get missed. Analysts burn out answering the same questions repeatedly.
By creating a conversational interface backed by governed metrics, we’re giving teams the ability to explore data without breaking anything, and without breaking stride.
We’ve helped clients across industries modernize their BI and AI ecosystems, but the missing link is often usability. That’s where we focus: building practical, scalable tools that make data work for humans, not just machines.
What’s Next
We’ve already started exploring how this same approach can be adapted for different environments – Teams instead of Slack, Snowflake instead of Databricks, or custom front-ends tailored to specific industries. The architecture is modular, and the use case is universal: give people governed access to data, without the friction.
DistillGenie is just one example of how natural language interfaces can unlock more value from the investments companies are already making in AI and analytics. And while this project was custom-built, it’s part of a broader shift we’re seeing, where the real innovation isn’t in building more tools, but in making existing ones actually usable across teams.
If this resonates, or if you’re exploring how to make your data more accessible, take a look at this walkthrough and demo, or reach out to us to start a conversation.