Orply.

ChatGPT’s Data Plugin Turns Questions Into Charts, Dashboards, and Reports

OpenAIFriday, November 20, 20264 min read

Alistair Gillespie presents ChatGPT’s Data plugin as a way to turn questions about data into different kinds of working outputs: a chart for a defined comparison, a dashboard for reviewing related measures, or an interactive report that combines evidence and narrative. The examples show it drawing on a Google Drive sales file and presenting capacity and flood data; Gillespie says it can also work with files, APIs and custom MCPs, and support tasks such as data-quality checks, KPI definition and sharing findings.

A question-and-chart response answers a bounded comparison

The sales example shows the Data plugin handling a focused question with a visual answer: using “Acme Sales 2026” in Google Drive, the user asks how sales changed from Q2 to Q3 2026 and requests a bar chart of each segment’s contribution. The response reports an increase of $923,452, or 20.5%, from $4.50 million to $5.43 million, then breaks the change out by segment.

20.5%
reported increase in sales from Q2 to Q3 2026

Industrial manufacturing accounts for $608,879 of the reported growth and construction for $293,443; facilities services contributes $19,025, and fleet maintenance $2,104. The point of the chart is not just to repeat the two quarterly totals. It makes the distribution of the change visible: most of the listed contribution comes from industrial manufacturing and construction, while the other two segments account for much smaller amounts. That gives the user a way to inspect what sits behind the headline increase.

This kind of response suits a question with a defined comparison and a specific breakdown the user wants to see. It returns an answer and a visualization in the same exchange. The plugin listing describes its analysis as grounded in the user’s data and its visuals as interactive; users can begin with their own data or try a sample dashboard and report. The presenter, Alistair Gillespie, characterizes the basic interaction as asking a question and receiving an answer with a visualization.

A dashboard puts related measures together for review

A dashboard serves a different purpose in the capacity example. The user asks which regions are running low on compute and asks for a dashboard for a capacity review. The resulting view groups fleet load and capacity with demand over time, alongside measures including 37,000 active demand, 89.1% fleet utilization, and 4,500 available slots. It also includes usable capacity, a failover limit, and queued workloads.

Rather than presenting only a response to the question, this view places several measures together in a format intended for review. That distinction matters when the task is not simply to learn one result, but to look across related indicators: the dashboard puts demand, utilization, available capacity, and workload measures in one place. Gillespie says users can build and share dashboards with their team using sites and familiar BI tools. The plugin’s skills listing describes dashboards as tools for monitoring performance, exploring drivers, and acting on metrics.

The example does not specify how often the dashboard updates or how a team uses it after sharing. What it does show is the intended format: a question about capacity becomes a view organized around a review, rather than a single answer or chart.

An interactive report can carry a data story

The flood example uses another form: an interactive report that combines narrative and visualization. Its headline says New York’s flood threat reaches beyond the waterfront. The accompanying text describes water at Carroll Street and Fourth Avenue in Brooklyn, nearly three feet deep by about 9:48 p.m. on September 1, 2021, after reaching that height in 21 minutes. An interactive 3D map shows flood depth across city blocks over time.

This is not the same job as the sales chart or capacity dashboard. The chart breaks down a particular change; the dashboard groups measures for a review; the report uses text and an interactive map to present an event as a story. Gillespie offers it as an example of telling a data story through an interactive report. The plugin listing also includes a skill for building reports from reviewed evidence.

The workflow can draw on existing tools and continue beyond an answer

Gillespie says the Data plugin works with files, APIs, and custom MCPs, and describes bringing data into ChatGPT as open-ended. The marketplace screen shows Data alongside tools including Hex, ThoughtSpot Spotter, PostHog, Omni, BigQuery, Amplitude, Microsoft Power BI, and Mixpanel. He says the list is not exhaustive and frames the plugin as compatible with tools people already use. The source does not detail the setup for each connection, but it presents the choice of data source as broader than the examples shown in the marketplace.

The listed skills extend the workflow beyond asking an individual question. They include assessing data quality, building dashboards and reports, converting an existing Data app into a document or slide deck, customizing reusable settings, and defining KPIs, driver metrics, guardrails, and scorecards. Gillespie also names product and business analysis and market sizing as examples of supported work.

After an answer, users can ask ChatGPT to propose next steps, update documents with new data automatically, or share findings through Slack, Teams, or email, he says. Taken together, the examples distinguish three useful outputs: a chart for examining a defined comparison, a dashboard for bringing measures together in a review, and a report for presenting evidence as a narrative. To get started, users can search for Data in the plugins marketplace and review its listing, which offers a sample dashboard and report as well as the option to begin with their own data. Availability depends on workspace settings.

The frontier, in your inbox tomorrow at 08:00.

Sign up free. Pick the industry Briefs you want. Tomorrow morning, they land. No credit card.

Sign up free