The AI Scenario Generator uses the OpenAI API to produce a draft Crisis Simulation scenario from a few inputs — you review and edit it before publishing to your catalog.
How to generate a scenario
- Navigate to Exercise > Crisis Simulations, and then click Create Scenario.
- From the Create your scenario options, click Create with AI.
The Create with AI window appears. - Fill in the required fields:
- Scenario Title
- Industry affected: Energy and Infrastructure, Financial Services, Government, Healthcare, Logistics and Supply Chain, Manufacturing, Other, or Technology.
- Type of attack (Attack vector): Data Breach, Denial of Service, Insider Threat, Malicious Code, Other, Phishing, Ransomware, Remote Working, Supply Chain Compromise, Targeted Attack, Unauthorised Access, Vulnerability Disclosure/Reporting, or Zero-Day Exploit.
Who is the threat? (Threat actor): Criminal Groups, Current Employees, Customers, Disgruntled/Former Employees, Human Error, Nation State, Other, Political/Social Activists, Red Hat Hackers, or Terrorist Groups.
-
(Optional) You can specify a Briefing prompt to give additional information about the scenario you want to create to increase the relevance to your organization and the outcomes you are trying to achieve. For further details, See What to write in the Briefing prompt below.
(Optional) In the Reference file section, you can click Upload or drag and drop a PDF to add additional context. For further details, see What documents should I upload in the Reference file field? below.
- Click Generate briefing. The scenario briefing appears. You can click Edit briefing to revise it, or accept it as-is.
- Click Generate scenario.
-
When the message Your scenario is ready for review appears, you can either click Preview Scenario or click Continue editing in builder.
From the builder you can edit, review, and publish the scenario the same way you would any custom scenario. See Article 6: Create a Custom Scenario for full details on the Scenario Builder.
What to write in the Briefing prompt
The Briefing prompt lets you give the AI more information about the scenario you want to create, so the output is more relevant to your organization and the outcomes you're trying to achieve. A well-written prompt makes a significant difference to the quality of the result.
Tips for effective prompting:
- Instruct the AI to adopt a specific persona — for example, "Act as a Crisis Incident Expert designing an ongoing crisis simulation exercise."
- Reference your organizational context: industry, size, the kinds of decisions your teams typically face.
- Be specific about your learning objectives and the audience.
- Include examples of the tone or style you want if you have them.
Using the Reference file field
You can upload a PDF — for example, a news article or report on a cyber attack relevant to your industry — to give the AI additional context when generating your scenario. PDFs are not stored after generation, and won't be used by OpenAI to train its models.
You're responsible for any documents you upload. Don't upload content containing sensitive, confidential, or personal information you're not authorized to share with a third-party AI system.
Reviewing and editing the output
Always review the AI-generated output before publishing. The generator currently produces text content only, so you may want to add images, videos, or other rich media. We also recommend checking the following in edit mode:
- Text formatting and narrative content
- That injects and options link logically
- That you're satisfied with the role
- Feedback and performance indicators — these aren't generated automatically and need to be enabled and configured
- Option rankings (great, good, okay, weak) — if you want ranked response data, you'll need to select the ranked options setting and assign a rank to each option
- Response confidence and justification settings
The outputs are generated by artificial intelligence and aren't verified by Immersive for accuracy. They don't represent Immersive's views, and Immersive makes no warranty as to the accuracy, completeness, or reliability of any outputs.
Scenario sharing and settings
If your organization has enabled Crisis Sim — Share custom scenarios for enhanced AI responses, the AI will draw on your organization's previously published custom scenarios to make future generations more relevant to your context. Scenarios are not shared externally — they're used only within your organization's generation requests.
Without Scenario Sharing, the AI uses only the title, dropdown selections, Briefing prompt, and any Reference file you provide. The output will be more generic, but you can edit it into shape just like a catalog scenario.
Organization admins can manage these settings at Settings > Crisis Simulations, in the AI section of the Platform Settings page.
The available configurations are:
- AI access with document upload and free-text interaction, but without Scenario Sharing (default)
- AI access with document upload, free-text interaction, and Scenario Sharing enabled
- No AI access
When Scenario Sharing is enabled, the data shared from your custom scenarios includes scenario title, description, industry sector, attack vector, threat actor, and inject and option titles and descriptions. Feedback, exercise information, reporting, account or organization information fields, and metadata are not shared. No data is used for model training, and shared data exists only in the temporary context window during generation.
OpenAI's servers are located in the US. All AI service data is stored in the Immersive US region. If you enable Scenario Sharing, scenario data is stored there until you switch the option off, at which point it's deleted.