Creating dashboards

Mammoth dashboards are built through a conversation with AI. You describe the charts and insights you need—the AI reads your data's structure and generates a live, interactive dashboard. No coding or manual chart configuration required.

Before you start

You can start from your data, a template, or an existing report:

• Your data needs a View, the prepared version of your data in a Mammoth project. If you haven't created one yet, set one up in your project first.
• Start blank and templates don't need data up front. A template comes with sample data, and you must connect your own data (Use my data) before you can publish it.

Note: Dashboards can only use data from projects you can access. You can't link Views from different workspaces.

Step 1: Choose where to start

1. Click Dashboards in the left sidebar
2. Click Create
3. On What do you want to build?, pick a lane

With my data shows your recent Views. Choose one, or click Browse all data to find any View you can access. You can also click Start blank to build from an empty canvas.

With a shape starts from something that already exists:

• A template—peek at one, or click Browse all templates
• Bring a report over—accepts PDF, Screenshot, Power BI, and Tableau
• Copy one of mine—start from a dashboard you already have

The AI reads your View's columns and sample data to understand what's available.

Step 2: Choose a format and describe it

After you choose data, pick a format card:

• Dashboard—tiles you come back to
• Doc—written findings, in order
• Slides—a deck
• Q&A—a page where you ask questions and the answers become cards

The header shows the View you're building from (for example "from Sales") with a Change link to pick another one, plus its row and column counts.

For Dashboard, Doc and Slides, the AI suggests ideas based on your columns, each with a short description. Select an idea to see a Roughly this sketch of the layout, then click Build this dashboard (Build this doc or Build these slides for the other formats) to build it. Or type your own description in the prompt box, ...or describe the dashboard you want. If the AI can't read your columns, it asks you to describe what you want instead.

Q&A doesn't suggest ideas. It offers Open a notebook, where readers ask their own questions of the data and each answer comes back as a chart or a number.

Click Add context under the prompt to attach saved Context: definitions or goals, so answers use your terms.

Be as specific or as general as you like:

General:

"Give me a sales performance overview"

Specific:

"Show me monthly revenue as a line chart, top 10 products by units sold as a bar chart, and a KPI card with total transactions this quarter"

The AI uses your column types to choose charts:

• Date columns → used for time-series charts
• Categorical columns → used as chart dimensions or filters
• Numeric columns → used as metrics, such as sums, averages, or counts

Build from an idea or submit your description. A progress screen shows the build steps—Analyze your data, Design the layout, Build the charts, and Finalize the pages (the wording changes for docs, slides, and Q&A). Then the editor opens.

Templates and plan limits: a template lets you preview the layout on sample data. Samples can't be published until you connect your own data. How many dashboards you can publish and generate depends on your plan.

Step 3: Review and refine in the editor

The editor has a left panel with three tabs and a live preview of the dashboard as viewers will see it:

• AI Chat—Describe changes in plain language
• Design—Format, page structure, navigation, style, and signature (see Dashboard themes)
• Data—The sources behind the dashboard

The Add menu in the header adds a Chart, Filter, Page, Cover, Q&A page, or Explore page. The header also has rename, history, a Desktop/Tablet/Mobile preview switch, Edit layout, undo and redo, and Preview.

When a dashboard has a Q&A page, an Ask AI button appears in the bar above its pages. It jumps to the Q&A page, and it is hidden while you are already on that page.

Examples of follow-up requests in the chat:

• "Change the bar chart to a horizontal layout"
• "Add a filter for region"
• "Split this into two pages—one for sales, one for marketing"
• "Make it show only data from this year"

Each request updates the dashboard in real time. The AI preserves your previous choices unless you explicitly ask to change them.

Save, then publish

Saving is explicit. When you have unsaved edits the header shows Unsaved changes, and publishing is disabled until you save. After you save, use the Publish button to go live (see Publishing and sharing dashboards).

What charts can you add?

Chart type	Best for
Line	Trends over time
Bar	Comparing categories
Bar (H)	Rankings
Donut	Part-to-whole proportions
Scatter	Correlation between two metrics
KPI	Single key metric with label
Table	Detailed row-level data
Heatmap	Two-dimensional density
Sankey	Flow and distribution between categories

The chart picker offers many more, such as area, combo, funnel, gauge, histogram, box plot, radar, treemap, waterfall, and map.

After creation

From the editor you can:

• Publish and share the dashboard
• Set up data sync to keep charts current
• Return at any time and keep chatting with the AI to change it