### 1.1 What can I create with Celestory?
AI has unlocked content creation at an unprecedented scale — everyone can now generate text, images, audio, and video in minutes. But what do you do with all that power? Celestory lets you turn it into real products: AI apps with your own rules, your own data, and your own identity. Through the visual modeling of your expertise, you can train your AI to build internal tools for your organization or deploy experiences to the general public — without writing a single line of code.
The more content you give your AI, the more important it becomes to master key concepts and choose the right LLM provider. Celestory is designed to make this journey accessible to non-technical creators: it gives you direct control over models, prompts, and logic through a visual interface built for humans, not developers.
AI can be activated at two levels:
- As a creator, you use AI to generate and refine content in the editor (text, images, video, audio).
- For the end user (called a "player" in Celestory), AI can be embedded directly into the experience to enable infinite, dynamic generation — personalized responses, on-demand images, speaking avatars — all triggered in real time during gameplay.
Enjoy maximum immersion through numerous formats: 📱 AI Apps, 💬 AI Chatbots, 📚 Choose Your Own Adventure stories, 🖼️ Visual Novels, 🎴 Cards, 🚪 Escape games, 🎬 Interactive films, 👆🏻 Point & Click, ❓ Quizzes, 🏥 Serious Games…
AI Apps are web applications fully created with Celestory and AI, instantly deployable as a PWA (Progressive Web App) or embeddable via iFrame:
🌐 Website  ·  📝 Blog  ·  📋 Form  ·  🤝 Meeting  ·  🎤 Presentation  ·  📖 Catalog  ·  🏠 User portal  ·  🎓 Online school  ·  📄 Documents  ·  🌍 Localisation  ·  📷 QR code  ·  🏅 Certification  ·  📊 Dashboard  ·  👁️ Data view
With No-Code and Vibe Coding powered by AI, the possibilities are truly infinite — any content, any tool, any experience you can imagine can be built and deployed without writing a single line of code.
Feel free to discover the community's creations and start your first project in a few minutes.
### 1.2 What is a sovereign AI model?
A sovereign model is an AI model developed by a European company (e.g., Mistral AI, based in Paris 🇫🇷) or in open source (e.g., Llama by Meta, Flux by Black Forest Labs), independent of the large American tech platforms.
Main advantages: they can be downloaded and run locally on your own servers, without sending your data to a third party — ideal for legal, medical, or corporate use cases requiring maximum data confidentiality.
For organizations requiring full data sovereignty, Celestory offers the Celestory AI Cube — a nano data center solution that lets you host Celestory itself, your AI apps, and your LLM models on your own infrastructure. Available as a Docker image, it can be deployed on-premise or in a private cloud with a single command. Contact us to learn more.
### 1.3 Which text AI models are available in Celestory?
Celestory gives access to the following providers directly from the editor:
Mistral and Llama (Meta) models are sovereign/open-source and can be downloaded for local execution after validation.
### 1.4 What is Vibe Coding?
Vibe Coding is the ability to iteratively create or modify an application by conversing with an artificial intelligence. Instead of manually writing programming languages (HTML, CSS, JavaScript), you give highly specific textual instructions (or "prompts") to the AI.
The AI instantly generates and updates the logic and interface of your application. In Celestory, you can embed this generated code within HTML5 blocks to create custom tools, such as asking the AI to develop a card mini-game or a full-featured "To-Do list".
### 1.5 What is no-code?
No-code means creating software and applications without writing programming languages. Instead of typing code (JavaScript, Python, HTML…), you build everything visually: assembling blocks, drawing connections, filling forms.
In Celestory, everything is built in the Graph — a visual flowchart where you drag, connect and configure blocks to describe the logic of your experience. No developer knowledge is required. Even complex features like AI integration, database connections, or conditional branching are handled by filling in text fields.
No-code makes digital creation accessible to everyone: teachers, trainers, marketers, entrepreneurs — no matter your technical background.
### 1.6 What is the AI Generation menu in Celestory?
The AI Generation menu is accessible directly from the Celestory editor and is organized into 5 tabs:
A quota counter at the top of each tab shows your AI credit consumption in real time (e.g., 45 / 200 generations).
### 1.7 How do I use the Text tab to generate content with AI?
The Text tab offers an integrated conversational chat to interact with language models (LLMs). You enter a message, optionally attach text variables from your project, and the model generates a response rendered in Markdown in real time.
Key features include: model selection, a token counter (showing history usage vs. context window), input estimation before sending, and the ability to attach project variables (style guides, game rules, methodology) to the prompt in one click.
### 1.8 How do I generate images with AI in Celestory?
The Image tab offers two modes:
- Text to Image: Describe the image you want (style, colors, atmosphere) and the model generates it from scratch. Supported models include Nano Banana, Flux 2 Klein 4B, and GPT Image 1.5.
- Image to Image: Provide one or more reference images and a prompt describing the changes. The model preserves structure and style while applying your instructions. GPT Image 1.5 also supports inpainting (retouching a specific masked area only).
### 1.9 What can I create with the Video tab?
The Video tab offers 6 generation modes:
- Text to video: Generate a synthetic video from a text prompt (LTX-2, Veo 3.1, Kling, Wan).
- Image to video: Animate a still image based on a movement prompt.
- Start and end frame: Provide a start image and an end image; the AI generates all intermediate frames.
- Image + Audio (AI Avatar): Synchronize lip movements of a portrait image with an audio file to create a speaking avatar.
- Video to video: Extend or modify an existing video.
- Motion Control: Apply the movements of a reference video to a character image (Kling v2.6).
### 1.10 What does the Audio tab allow me to do?
The Audio tab (powered by ElevenLabs) offers 4 modes:
- Text to Speech (TTS): Generate voice-overs from text with 20 available voices, with control over stability, similarity, and speed.
- Music: Generate an original music track from a text description of the style and atmosphere.
- Sound Effects: Generate tailor-made sounds (button clicks, ambient environments, action effects) with adjustable duration and loop option.
- Multi-voice Dialogue: Compose a conversation between multiple characters, each assigned a different voice, exported as a single continuous audio file.
### 1.11 What is a token and why does it matter in Celestory?
A token is the basic unit used by AI models to process text. One token corresponds approximately to ¾ of a word in English (e.g., "hello" = 1 token, a 10-word sentence ≈ 13 tokens).
Each model has a context window — the maximum number of tokens it can process in a single conversation. Celestory displays the token count in real time (history vs. context window, e.g., 12,530 / 256,000 tokens) so you can optimize your prompts and avoid exceeding the model's memory.
### 1.12 What is a competency prompt and how do I use it?
A competency prompt (also called System Prompt or Role) is a reference document — style guide, methodology, business rules — that you attach to your AI requests to guide generation. For example:
- For Vibe Coding: attach your HTML/CSS style guide as a variable so the AI respects your colors and typography.
- For document generation: attach a Markdown methodology so the AI follows your structure conventions.
In Celestory, you store these guides as text variables in your project and inject them into any AI conversation with one click.
### 1.13 What are the in-game AI blocks available to players?
Two blocks allow players to trigger AI generations during play:
- 🟦 AI text request block: Connects your scenario to a language model. Inputs include Prompt, Role (System Prompt), History, Model, Temperature, and optional Images. Outputs include the Generated Response, a Success signal, and an Error signal.
- 🟦 AI image request block: Transforms a text description into a dynamically generated image. Inputs include Prompt, Model, optional reference Image, and Negative Prompt. Outputs include Image URL(s), Success, and Error.
### 1.14 How do I manage AI credit consumption in player experiences?
To control costs when players trigger AI requests:
1. Limit requests per session, per player, or per scene.
1. Use lightweight models (ministral-3b, Nano Banana) for frequent interactions; reserve premium models for key moments.
1. Monitor consumption in the Previous generations tab (filterable by "in-game" source).
1. Handle errors gracefully with the Error output of AI blocks to provide fallback behaviors (pre-defined responses, default images).
1. Use strict System Prompts to constrain AI behavior and reduce unnecessary tokens.
### 1.15 What is the difference between a negative prompt and a system prompt?
- A Negative Prompt tells the AI what it must not include in the generation (mainly used for image and video). Example: "no blur, no deformed hands, no text".
- A System Prompt (or Role / Competency Prompt) sets the AI's personality and behavior rules before a conversation. Example: "You are a sarcastic wizard who only answers in verse". It is used for text generation to maintain a consistent character throughout the conversation.
### 1.16 How do I view and reuse my past AI generations?
The Previous generations tab gives access to the complete history of all your AI generations (text, image, video, audio). You can filter by type (text, image, video, audio) and source (editor or in-game), paginate through older generations, and reuse previously generated outputs without regenerating them.
### 1.17 What is a serious game?
A serious game is a "video game" with a serious purpose that allows learning through an application that digitizes real-world logic. Celestory allows you to create more than simulations — real experiences that assist each person in your audience to train or perform actions, such as accessing private data you share with them (documents, recommendation lists, etc.) and controlling the software you give them access to (email, drive, CRM, etc.).
### 1.18 What are the differences between a serious game and e-learning training?
E-learning is a format closer to a linear presentation, usually with a quiz for evaluation. A serious game is more immersive. It contains content (which can also be played linearly), but also offers an experience that is personalized according to the user's actions: new scenario branches, personalized AI responses, profiling with numerous scores such as rankings, badges, or challenges. Finally, a serious game can have different formats: interactive presentation, chatbot, interactive comic strip, cards, etc.
### 1.19 What is the difference between an escape game and a serious game?
An escape game is a game where the principle is to find the exit from a locked place. It can be transformed into a serious game if there is a pedagogical aim. It can be designed autonomously or in "phygital" mode (physical-digital), i.e., a digital experience in addition to a physical one (e.g., scenarios, clues, objects in digital...).
### 1.20 What are the advantages of serious games?
First of all, because of the game aspect, it is an attractive experience that puts the user in a situation.
Gamification maximizes user engagement through a reward system.
A serious game can save the user's progress online through different sessions. Connecting to a relational database allows, for example, to offer multi-user experiences.
The serious game allows personalization of the experience through the memory of the user's actions but also enables real-time study of the multiple factors that explain player behavior.
### 1.21 Can we make games on any subject?
This is the beauty of Celestory — you can create games on all the subjects you want, including the most "niche" where no game exists yet, thanks to the accessibility of creation.
You can create your game and download it to distribute it wherever you want. By hosting it publicly on Celestory, you must ensure that you have all the operating rights and that the content is entirely legal.
### 1.22 Why make a serious game?
To raise awareness among your audience, while monitoring the multi-causality that explains the different approaches by that audience, allowing you to confirm or identify categories of behavior.
### 1.23 What are the advantages of Celestory? Can I create alone or in a team?
A Celestory serious game is the guarantee of having an experience that is easy to evolve visually without coding ("no-code"), alone or in a team. Your project will be easily updated by accessing the data you share with it. You can integrate AI capabilities for infinite queries, and you can even integrate it into many software programs to automate your work (send emails, collect data, summarize documents...).
Compatible with all devices and deployable instantly, Celestory technology is also the best way to create an immersive "digital product", which you can monetize if you wish.
### 1.24 Does Celestory offer team collaboration features?
Yes, Celestory allows you to grant visibility or modification access to your game project to members of your team. It is highly recommended to share your projects in read-only mode to prevent collaborators from accidentally modifying the Graph (scenario).
Your collaborators can then Duplicate the shared project from their dashboard to their own "My projects" space, allowing them to work on their local copy safely without ever affecting the original creator's version.
### 1.25 Can we customize the Celestory interface and the games created with our own graphic charter and branding?
Yes, Celestory offers extensive visual customization. You can modify colors, fonts, background images, and the overall style of your experience from the Module Settings > Application Style panel. You can also upload your own logo, apply custom CSS, and build custom UI components (ATH menus) to fully reflect your brand identity.
### 1.26 How does Celestory handle translation into different languages?
Celestory supports more than 200 languages and offers complete tracking of the number of blocks you have translated and how many you have left to translate. Celestory generates a PO file that can then be pre-translated with AI.
### 1.27 What technology is the Celestory engine based on?
The Celestory engine is mainly developed in React. Games are created in HTML5.
---