The Data Q&A view allows you to explore data by asking questions in natural language and receiving structured, explainable visual answers directly inside an Omniscope report.

In this article, we'll walk through a practical example using public climate and energy data. We'll explain how the Data Q&A view works, and show how it can be configured for different audiences - from analysts to report consumers.

What is the Data Q&A View?

The Data Q&A view is a special type of view in Omniscope. Instead of configuring the chart up front, you ask a question in plain English and receive a structured answer that can include:

• Headline summaries
• Badges explaining assumptions and limitations.
• Charts (KPIs, bars, lines, pies, tables)
• Clear explanations and caveats.

Importantly, the Data Q&A view:

• Can see and query all data sources connected to the report.
• Explains how it arrived at an answer.
• Lets you control how much of the reasoning is visible to an end user.

The datasets used in this example

In this example, we use three public datasets related to climate, energy and transport. They are related but not perfectly aligned - which makes them a good real-world test case.

CO2 emissions data.

The CO2 dataset comes from Our World In Data, a research organisation based at Oxford University. This dataset includes emissions by country and year and per-capita metrics.

Source: 

https://ourworldindata.org/co2-and-greenhouse-gas-emissions?utm_source=chatgpt.com

Global energy data

The global energy data also comes from Our World in Data. It includes electricity generation by source (coal, gas, oil, renewables, etc.), along with energy consumption and share metrics.

Source:

https://ourworldindata.org/energy

Electric vehicle adoption data

We also include EV sales data from the International Energy Agency (IEA). This dataset shows electric vehicle sales by country and year.

It allows us to explore questions such as:

Are high-emission countries also transitioning to EVs

Create a new Omniscope project and drop each of these CSV files onto the workflow. 

Working with the Data Q&A view

Before you can use the Data Q&A view you need to enable and configure AI in Omniscope. Full instructions can be found here.

Open the Add Block menu and add a Data Q&A report to the workflow.

Now connect all three source blocks to the report block. Click on the report block to open it. You should see the Data Q&A view inside the report. Alternatively you can add a Data Q&A view to any existing report by clicking the Add view button.

The Data Q&A view consists of two main areas:

1. A chat panel on the right, where you ask questions
2. A results panel on the left, where answers are displayed.

Unlike many other views, the Data Q&A view is not tied to a single dataset in the report data model (which could be joined from multiple report inputs). Instead it is automatically linked to all report inputs, and can reason across all data sources available in the report.

Data Q&A view configuration

You can configure how the Data Q&A view behaves by opening the chat settings. Here you can choose:

• The model
• The thinking level
• The verbosity of responses

In this example, we use GPT 5.2 with low thinking and medium verbosity. This combination produces answers that are fast and direct, while still being detailed enough to understand without overwhelming the viewer.

By default the Data Q&A view starts by asking:

Tell me about my data

The AI typically responds with a short summary of each dataset connected to the report. This makes it immediately clear what data is available and what the AI can analyse. It can also provide a series of options for analysis, rather than manually typing in a question.

The initial behaviour of the Data Q&A view is fully configurable. An editor can choose how (or if) the AI is invoked when a new session starts.

Open the view settings and expand the Q&A section. 

Here we can configure:

• Mode.
  Choose between showing the answer with a chat sidebar, or an answer-only view.
• Custom prompt
  A system prompt that tailors the AIs behaviour - for example, encouraging more visual answers or adding domain specific context.
• Initial message behaviour
  This determines what happens the view loads:
  • Inherit from integration - uses the default behaviour, which can also be configured globally by an administrator
  • Hidden message - a system message is sent automatically and only the answer is shown
  • Assistant message - displays an introductory or suggested message without invoking the AI until the user asks a question
  • User message - sends a visible user question immediately and shows both the question and the answer

These options make it easy to adapt the view for different audiences and use cases.

Asking our first question

Lets start with our first question:

Which countries are the worst CO2 emitters per person

Enter the question in the chat panel and hit return. You'll see a brief period of thinking, followed by the answer streaming into the results panel.

Once the answer has finished streaming we can examine it in more detail.

The Question and Answer sections are expanded by default.

At the top of the answer we see a clear headline summarising the result.

Below the headline are a number of badges. Each badge represents an assumption or limitation used by the AI. Clicking a badge reveals more detail.

The main body of the answer is a mix of text and charts. Using the tools menu on any chart, we can download the underlying data or export the chart as an image.

Above the answer the Question section shows the user question and the AI's interpretation of that question.

The Activity section shows the thinking and the queries the AI executed. Each query can be opened to display the results in a table, and - like other charts inside the Data Q&A view - data or an image can be downloaded.

The Explanation section shows how the AI reached its conclusion. Each pill represents a step in the reasoning process, and clicking a step reveals the queries used.

The Activity/Explanation sections are important because they show:

• What the AI actually did
• That the answer is based only on the report's data
• That no external data sources were used

As an editor you can decide which sections are visible. If you want to show just the answer you can:

1. Open the view settings
2. Open the Result section
3. Untick the sections you want to hide

Asking further questions

Lets continue with two more questions.

First:

Which countries are seeing the fastest growth in electric vehicle adoption

In this answer, we can see that some smaller countries show very rapid growth (for example, Bulgaria and Costa Rica). This kind of insight is difficult to spot in a static dashboard.

Next:

Are the countries adopting electric vehicles faster the ones with lower CO2 emissions per capita?

Here, the AI considers all three datasets together.

In some cases the AI may explain it cannot reliably provide answer, for example because of differences in data coverage across the datasets. Rather than hiding this, the Data Q&A view makes the limitation explicit, showing where the data is strong and where it is incomplete.

Conclusion

The Data Q&A view doesn't replace your dashboards - it complements them. 

It helps you and your users explore data, understand trade-offs and explain results in a way that's transparent and repeatable. Because it show its workings, its something analysts can trust.

If you're already using Omniscope, the Data Q&A view offers a new way to interact with your data - one that sits naturally alongside your existing reports.