Design Agent is an AI-powered capability that streamlines the creation and customization of campaign templates. It enables teams to generate on-brand campaign designs rapidly without coding, significantly reducing production time and dependency on design and development resources.

Design Agent supports two primary modes of operation:
Pop-Up Mode
Used for creating standard pop-up designs that may include images, buttons, countdown timers, static coupon codes, and text elements.
Gamification Mode
Used for generating interactive, playable campaign experiences through natural-language prompts, such as Rock–Paper–Scissors, Tic Tac Toe, Word Guess, and Digit Recall.
These two modes enable marketers to design engaging, high-impact campaign experiences while preserving brand consistency and minimizing operational complexity.
Design Agent also marks an advancement toward AI-driven marketing personalization, strengthening automation capabilities and interactive campaign design within OnSite products.
Pop-Up Mode
Pop-Up Mode is designed to create pop-up designs featuring images, buttons, countdown timers, static coupon codes, and text elements using generative AI.
Capabilities
Generates ready-to-use pop-up layouts based on natural-language prompts.
Allows users to edit or regenerate designs using additional prompts.
Supports manual fine-tuning through the in-editor floating menu.
Key benefits
Reduces popup creation time from hours to minutes.
Minimizes reliance on design and technical teams.
Improves campaign speed, quality, and consistency across teams.
Gamification Mode
Gamification Mode automates the creation of interactive, game-based campaign experiences designed to increase user engagement and conversion.
Capabilities
Generates five complete, playable game types in its initial release:
Rock–Paper–Scissors
Tic Tac Toe
Color Hunt
Word Guess
Digit Recall
Additional capabilities include:
Creating visual and gameplay variations aligned with brand identity
Reusing built-in logic for scoring, animation, and interactivity
Enabling scalable self-service game creation for partners and internal teams
Key benefits
Accelerates expansion of the gamification library through AI-generated templates.
Reduces dependency on external teams by up to 90%.
Enables partners to react immediately to seasonal, campaign, or trending concepts.
How Design Agent works
Navigate to Smart Design Creator from the Onsite Experiment’s Design step using the top menu.

A built-in help and education panel is available within the interface, where you can review capabilities, limitations, and important usage notes to better understand how Design Agent works.

You can choose either Pop-Up or Gamification mode.

Pop-up Mode: When selected, the system displays prompt suggestions (e.g., “Create a seasonal signup popup with a countdown timer”) to help you start the chat and generate a design instantly.

Gamification Mode: When selected, the system displays a list of available game templates (e.g., Rock–Paper–Scissors, Tic Tac Toe, Word Guess). You must pick one game type to continue customizing it through prompt instructions or direct edits.

Navigation Mode, particularly valuable in Gamification Mode, allows you to switch between different game screens or states (for example, start screen, gameplay, win screen, loss screen). This enables you to preview each step of the experience and fine-tune the design and content for every stage of the game.

You can describe what you want using natural language – e.g.:
“Create a holiday-themed popup with a countdown.”
“Generate a pastel-colored Tic Tac Toe game for a fashion brand.”
The system generates a fully functional design for a pop-up or game template. Both layout and logic (e.g., game scoring, animations, interactivity) are automatically created.
You can instantly undo or redo recent changes from the top bar, ensuring flexibility and creative freedom during customization.
You can click directly on elements to modify visuals, copy, or layout. The floating editor menu enables fine-tuning without re-running generation.

At any time, you can start a new design session using the Create a New Chat option, which clears the previous prompt and launches a fresh AI-assisted design process.
Once finalized, save and publish your design through the standard campaign workflow (segment, rules, launch).
Limitations and error messages
The Design Agent uses AI to interpret user prompts and generate campaign designs. However, certain inputs, conditions, or technical issues can result in limited or failed outputs. Below are the main cases and the messages you may encounter:
Successful scenarios
Type | Description | System Response |
|---|---|---|
Successful Generation | The prompt is valid, supported, and clearly structured. | |
Semi-successful Generation | The AI partially understands the request but omits minor details (e.g., missing styling or limited animation). | The design is generated using partial elements; you are advised to refine the prompt for better results. |
Prompt-related failures
Type | Description | System Response |
|---|---|---|
Irrelevant Prompt | You enter a prompt unrelated to design or gamification (e.g., “How to be rich?”). | “Design Agent can only help you create designs. You can enter another prompt to start creating.” |
Unsupported Game | You request a game type not among the five supported during initial development (e.g., “Create a Flappy Bird game”). | “Design Agent can only create gamification designs based on the available game types. Please enter another prompt and try again.” |
Not Understood | Your prompt is ambiguous or incomplete, and the AI cannot infer a valid design. | “Your request couldn’t be applied. You can rephrase your prompt and try again.” |
System-related failures
Type | Description | System Response |
|---|---|---|
Backend Error / Timeout | A server issue or long generation time prevents design creation. | “Your design couldn’t be created right now. You can wait for a while and try again.” (Includes a “Try Again” button.) |
Failed Generation | The system fails to produce a valid output due to an internal error or unexpected prompt structure. | “Generation failed. Please refine your prompt or try again later.” |