For years, the data team's job description could be summarized in one sentence: translate business questions into SQL and return the results.
That translation layer is exactly what AI is automating. And if you're a data professional reading this, that might sound threatening. It's not — but it does mean the job is changing faster than most org charts reflect.
What's actually being automated
Let's be specific about what AI-native tools handle well today:
- Ad-hoc queries — "How many signups did we get from paid search last week?" No ticket, no queue, no two-day turnaround.
- Data exploration — "Show me everything we know about customers who churned in their first 30 days." Pattern finding at scale.
- Report generation — Weekly business reviews, board decks, metric summaries. The scaffolding, not the insight.
What's not being automated — and won't be for a long time:
- Defining what to measure — Choosing the right KPIs requires domain knowledge and organizational context that no model has.
- Data governance — Who can see what, what's the source of truth, how do we handle PII. These are human decisions with legal and ethical weight.
- Strategic interpretation — "Revenue is down 12%" is a fact. "We should pivot our enterprise motion because mid-market is showing stronger unit economics" is a judgment call.
The analyst becomes the architect
The data professionals who thrive in this shift will move up the stack. Instead of spending 60% of their time writing queries and 40% on analysis, that ratio flips — and then some.
Their new responsibilities look more like:
Curating the intelligence layer
Someone needs to define which data sources matter, how they connect, what business logic applies, and what "good" looks like. This is the foundation that makes AI-generated answers trustworthy. Garbage in, garbage out — the cliché exists because it's true.
Teaching the organization to ask better questions
The bottleneck shifts from "can we get this data?" to "are we asking the right questions?" Data teams become coaches — helping product, marketing, and leadership frame problems in ways that data can actually illuminate.
Building the guardrails
When anyone in the company can query anything, governance becomes critical. Data teams design the permissions, audit trails, and quality checks that keep self-serve analytics from becoming self-serve chaos.
A practical migration path
If you're leading a data team through this transition, here's a framework we've seen work:
- Audit the query queue. What questions come up repeatedly? Those are your first automation candidates.
- Deploy conversational BI for the long tail. Free your team from the ad-hoc requests that eat their calendar.
- Redirect capacity upward. Take the hours saved and invest them in data modeling, metric definitions, and stakeholder education.
- Measure differently. Track time-to-insight and decision velocity, not tickets closed or dashboards built.
The opportunity
The data team isn't shrinking — it's leveling up. The people who spent years becoming SQL wizards now get to spend that expertise on the problems that actually require human judgment.
That's a better job. And for the organizations they serve, it means data stops being a service desk and starts being a strategic advantage.
We're building Dan to be the tool that makes this transition possible — not by replacing data teams, but by giving them their time back.