Creating a Dashboard with AI

Overview

The AI Custom Dashboard feature lets you describe the dashboard you want in plain English and have Strobes AI build it for you. Instead of manually configuring each widget, you tell the AI what data matters — and it queries your instance, proposes a widget plan, and creates the dashboard on your behalf.

There is also a second AI option — AI Insight Report — which generates an auto-refreshing security insights report rather than a configurable dashboard. This article covers AI Custom Dashboard only.

Before You Start

• The AI pulls live data from your instance (findings, assets, custom fields) to tailor the widget plan. The more data in your workspace, the more accurate the recommendations.
• Review the proposed widget plan before confirming, you can request changes before anything is created.

Step 1: Open the Dashboards Page

Click Dashboards in the left sidebar. The page displays all your dashboards — default templates and any custom ones your team has created.

Step 2: Launch AI Custom Dashboard

Click the ✦ Create with button (labelled New! AI-powered) in the top right. A small dropdown opens with two options:

Click AI Custom Dashboard.

Step 3: Describe Your Dashboard

A text input dialog appears over the Dashboards page.

Type a natural language description of the dashboard you want. Be as specific as you like — mention the data you care about, chart types you prefer, or the audience the dashboard is for.

Example prompts:

• Build a dashboard with multiple widgets covering findings severity, asset types, and SLA breaches.
• Create an executive summary dashboard for the CISO with KPI metrics and risk distribution charts.
• Build a dashboard focused only on critical and high severity open vulnerabilities with donut charts and a pivot table.

When you're done, click the send button (→) to submit.

Step 4: The Strobes AI Interface

After submitting, the dialog closes and the Strobes AI interface opens — a dedicated chat environment where the AI builds your dashboard.

The interface has the following areas:

Top bar — shows "Strobes AI", a green Running status badge (changes to Connected when idle), the AI model in use, and action icons (download, close, etc.).

Left panel (Tasks) — shows the current run status and any queued tasks. During processing, the current run shows as "running".

Chat area — the conversation thread between you and the AI. Your prompt appears as the first message. The AI responds in real time, showing its thinking steps, tool calls, and responses as it works.

Status bar (bottom) — shows what the AI is currently doing (e.g., "Dashboard Builder is running a tool…").

Back button — click to return to the Dashboards page at any time.

Step 5: AI Discovers Your Data

The AI immediately begins querying your instance to understand what data is available. You'll see it "think" briefly and then call a series of tools.

The AI runs three data discovery calls:

• ListCustomFields — discovers your custom fields for compliance, risk scoring, escalation, etc.
• FetchStrobesBugs — pulls a sample of your findings to understand severity distribution, states, and volume.
• FetchStrobesAssets — pulls your asset inventory to understand asset types, sensitivity, and exposure.

Once complete, the AI presents a Data Summary and a proposed Widget Plan.

Step 6: Review the Widget Plan

The AI presents a summary of what it found in your instance, for example:

• 511,086 total findings — 62k Critical, 240k High, 177k Medium (464k SLA breached!)
• 18,374 assets — 12k Network, 561 Web, 195 Mobile, 17k exposed
• Extensive custom fields for compliance, risk scoring, escalation, and more

Below the data summary, the AI proposes a widget layout. A typical plan looks like this:

The AI also tells you what this layout gives you — for example: quick KPI metrics for executive summaries, risk distribution visibility (severity and state), asset inventory overview, and detailed vulnerability breakdown.

The AI then asks: "Would you like me to preview these widgets now, or would you prefer to adjust the plan first? (I can add/remove widgets, change chart types, add filters like 'only critical exploitable findings', etc.)"

Step 7: Confirm or Adjust the Plan

You have two choices at this stage:

Option A — Build it as-is Reply with something like yes make the dashboard live or go ahead and create it. The AI proceeds immediately to building all widget configurations.

Option B — Refine the plan first Request changes in plain English before anything is created. For example:

• Remove the asset widgets and add a line chart for new findings over time.
• Change the Findings by Severity from a donut to a bar chart.
• Add a filter so all widgets only show critical and exploitable findings.

The AI will revise the plan and ask again before building.

Step 8 — AI Builds the Dashboard

Once confirmed, the AI moves to the build phase. It announces: "Perfect! ✅ I'm proceeding to STEP 3 — PREVIEW. Let me create all 8 widget configs and show you the live dashboard in the chat."

Behind the scenes, the AI runs a sequence of tool calls:

1. GenerateRqlFilter — generates the query filter syntax for each filtered widget.
2. AiFilterResponse — validates and applies the filter expressions.
3. CreateWidgetConfig (×8) — creates each individual widget configuration, one per widget in the plan.

The left panel shows "Current Run — running" while this is in progress. Each tool call shows a spinning indicator while active, and a green checkmark when complete.

Step 9 — View the Completed Dashboard

Once all widget configs are created, the AI returns a preview of the completed dashboard in the chat. From here you can:

• Click View Dashboard (or the dashboard name) to open it full-screen in the Dashboards page.
• Continue the conversation to make further adjustments — for example, add a bar chart for new vs active findings over time or rename the dashboard to CISO Overview.
• Use the Back button to return to the Dashboards page directly.

The dashboard appears in your Dashboards grid alongside the default templates, labelled with its creation date. It can be filtered, exported, and shared just like any other dashboard.
Dashboard Created!

Tips

Be specific in your prompt. The more context you give, the better the widget plan. Mentioning the audience ("for the CISO"), the data focus ("only SLA-breached critical findings"), and preferred chart types ("I prefer donut charts over bar charts") helps the AI make choices that match your needs without extra back-and-forth.

Review the data summary before confirming. The AI shows you exactly what it found in your instance — findings count, severity breakdown, asset count, and asset types. If the numbers look unexpected, stop and investigate before building the dashboard.

Refine before building. It's faster to adjust the widget plan in the chat (before the AI calls CreateWidgetConfig) than to edit individual widgets after the dashboard is created. Use the conversation step to get the layout right first.

You can keep refining after creation. The Strobes AI chat stays open after the dashboard is built. You can continue the conversation to add widgets, change chart types, or apply filters — the AI will update the dashboard accordingly.

The AI uses your live data. The KPI values and chart data in the completed dashboard are pulled from your actual instance in real time, not sample data. What you see in the preview is accurate.