The Generate page lets you create secure, optimized cloud architectures using natural‑language prompts. Dawnguard guides you through the process, asking clarification questions, gathering relevant context, and producing a structured architecture design that aligns with your guardrails and organizational context.

Use Generate to:

Start a new architecture from scratch
Refine or extend an existing design
Ask questions about design decisions
Turn conversations into reusable, saved designs

Generate methods

Generate with chat

Open Generate to start a new chat.
Describe the system, workload, or requirements
Click ↑ to submit
Dawnguard asks for missing details when needed.
The system gathers requirements and generates the initial architecture.
A status indicator shows real‑time progress (e.g., gathering requirements, generating design).
You can stop generation at any time using the Stop button.

Generate from image

You can upload architecture diagrams, sketches, screenshots, or other visual references to provide additional context during generation.

Click the + button in the chat input area.
Select one or more supported image files.
Add any additional requirements or instructions in the chat.
Submit your request.

Dawnguard analyzes the uploaded image and uses the visual context to generate or refine the architecture.

Supported Image Formats: .png, .jpg, .jpeg, .gif, .webp

Upload limits:

Maximum file size: 4 MB per image
Maximum of 5 images per chat message

Generate from document

You can upload architectural and technical documentation to provide additional context during generation.

Examples include:

Architecture design documents
System descriptions
Requirement specifications
Technical documentation

Click the + button in the chat input area.
Select one or more supported document files.
Add any additional requirements or instructions in the chat.
Submit your request.

Dawnguard analyzes the uploaded documents and uses the content to generate or refine the architecture.

Supported document formats: .pdf, .docx

Upload limits:

Maximum file size: 4 MB per document
Maximum of 5 documents per chat message

Generate from code

You can upload Terraform files to generate architectures from existing or planned cloud environments.

This helps you:

Visualize infrastructure from code
Validate architecture designs
Document existing cloud environments
Review Terraform configurations

Click the + button in the chat input area.
Select one or more supported code files.
Add any additional requirements or instructions in the chat.
Submit your request.

Dawnguard analyzes the uploaded code and uses it to generate or refine the architecture.

Supported code formats: .tf

Upload limits:

Maximum file size: 4 MB per file
Maximum of 5 files per chat message

Generate from Multiple Sources

You can combine supported images, documents, and code files in a single request to provide additional context during generation.

For example, you might upload:

An architecture diagram and supporting documentation
A Terraform file and architectural requirements
Screenshots, documents, and code related to the same environment

Dawnguard uses the provided context to generate or refine the architecture design.

Example Prompts

Use clear prompts to describe what you want Dawnguard to generate, review, or improve.

Examples:

Design a new architecture
“I want to design a system that achieves [goal] on Azure.”

Design for a specific cloud provider
“I want to design a system that achieves [goal] on AWS.”

Review an uploaded design
“I want to check this design for gaps against my guardrails.”

Improve an existing design
“I want to add a service to an existing design.”

You can include additional context in your prompt, such as the cloud provider, workload requirements, security needs, or files you have uploaded.

Dawnguard uses your prompt and any uploaded context to generate, review, or refine the architecture design.

‍

Review and refine

Refinements made in chat after saving will not update the saved design
You can continue the chat conversation to explore changes, adjust requirements, or generate an improved version, but this will produce a new generated design, not modify the saved one
You can always refine in chat, but saved designs never change
Once you save a design, it becomes a static snapshot

After generation, you can use the chat to review and refine the architecture.
You can continue iterating as long as the design has not been saved yet.

You can:

Ask why specific components were chosen
Modify requirements or constraints
Explore alternatives
Regenerate sections of the design
Keep iterating until you’re satisfied

Save as design

When you’re ready to preserve the generated design:

Open the generated architecture
Click Save as design

Your design is added to Designs for later access.

Once you save a design, it becomes a static snapshot.

‍

Chat history

Every generated design maintains its full chat history.

You can:

Review earlier reasoning
Continue the conversation after saving
Explore refinements or new directions
Generate new versions without altering saved designs

Provide feedback

Click Thumbs up or Thumbs down on chat responses to help us improve Dawnguard future chat results.

Copy text

You can click Copy on any message to extract:

Summaries
Component explanations
Design decisions
Architecture rationale

Perfect for documentation, reviews, and handovers.

Designs

Resources