OpenAI just handed non-technical workers a data analyst — and it lives inside ChatGPT Work, ready to turn raw company data into interactive dashboards using plain English.
What the Data Agent Actually Does
The new Data agent in ChatGPT Work lets you connect your company's data sources and ask questions the way you'd ask a colleague: "Show me last quarter's sales by region" or "Which customers churned after the price change?"
It doesn't just spit out a table. It builds interactive dashboards — charts you can filter, drill into, and share — without you writing a single line of SQL or touching a spreadsheet formula.
A Practical Workflow You Can Try Today
If your team is on ChatGPT Work, the entry point is straightforward: connect a data source (think Redshift, BigQuery, Databricks, Snowflake or similar systems, plus files from Google Drive or SharePoint), then describe the insight you need in natural language. The agent handles the query, the visualisation, and the layout.
A useful starting prompt: "Compare monthly active users from January to August and flag any months where growth dropped below 5%." That single sentence replaces writing a query and building the chart by hand.
The real productivity unlock is iteration speed — you can refine the dashboard by just talking to it, the same way you'd give feedback to a human analyst.
What This Means for Learners
This is a textbook example of why AI literacy is now a core workplace skill, not a nice-to-have. Understanding how to prompt an AI agent well — with the right context, the right constraints, and the right output format in mind — is what separates someone who gets a useful dashboard from someone who gets a confusing one.
If you want to get sharper at directing AI agents like this one, our AI Agents course breaks down exactly how these systems think and how to steer them effectively. And if you're curious about the engineering logic behind multi-step agent workflows, Loop Engineering with Claude gives you the mental model to understand what's happening under the hood.
The bottom line: the people who thrive with tools like this won't be the ones who know the most about data science. They'll be the ones who know how to ask the right questions.
