Story Mode: How Interactive AI Narratives Work
Story Mode lets you step into interactive narratives with your AI companion — not pre-written scripts, but stories that unfold dynamically based on the choices you make. This guide explains exactly how the system works under the hood.
---
What Story Mode Actually Is
Story Mode is a structured narrative experience where:
1. You choose a story template (setting and premise)
2. The AI generates a scene — descriptive text that sets the moment
3. You are presented with 3 choices for how to respond
4. Based on your choice, the AI generates the next scene
5. This continues until the story reaches a natural conclusion
The key difference from a regular conversation: Story Mode has narrative structure. There are scenes, dramatic tension, and consequences for your choices. The AI is not just chatting — it is building a story around your decisions.
---
Available Story Templates
We currently offer 6 story templates, each with a distinct setting, tone, and narrative arc:
Template Setting Tone
---------- --------- ------
Date Night An evening out together Romantic, warm, intimate
Adventure Exploration and discovery Exciting, playful, unpredictable
Mystery Something strange is happening Suspenseful, intriguing, atmospheric
Fantasy A world beyond reality Imaginative, epic, wonder-filled
Beach Getaway A coastal escape Relaxed, sensual, carefree
Midnight Encounter Late-night connection Intense, intimate, charged
Each template provides the AI with an initial narrative context — the setting, the mood, and the general direction of the story. But from there, your choices determine everything.
---
How Scenes Are Generated
Each scene is generated fresh by the language model. Here is what happens when you make a choice:
Step 1: Context Assembly
The system builds a prompt that includes:
- The story template (setting, tone, genre conventions)
- Your companion's personality (their character traits, speaking style, relationship to you)
- Your current relationship level (how intimate the narrative can be)
- All previous scenes in this session (narrative continuity)
- Your choice history (every decision you have made so far)
- Key narrative events (stored as structured JSON — plot points, revelations, emotional beats)
Step 2: LLM Generation
The language model receives all of this context and generates:
- Scene text: 2-4 paragraphs of descriptive narrative
- Three choices: Options A, B, and C for what happens next
Step 3: Choice Presentation
The three choices follow a consistent design philosophy:
- Choice A (Safe): The comfortable, expected option. Low risk, pleasant outcome.
- Choice B (Bold): The adventurous option. Takes the story in a more exciting direction. May introduce complications or surprises.
- Choice C (Premium/Daring): The most dramatic option. Costs additional credits. Leads to the most intense, unexpected, or emotionally charged narrative paths.
This is not random — the choice structure is part of the prompt instructions to the AI. We ask it to generate options that represent genuine narrative alternatives, not just slightly different phrasings of the same outcome.
---
Narrative Continuity
One of the hardest problems in AI storytelling is consistency. How do you make sure the AI remembers that you found a mysterious key in scene 3 when you are now in scene 8?
We solve this with a structured event tracking system:
Event Storage
After each scene, the system extracts key narrative events and stores them as structured JSON:
```
{
"scene": 5,
"events": [
{"type": "discovery", "detail": "Found a locked diary in the attic"},
{"type": "emotional", "detail": "Shared a vulnerable moment about fear of the dark"},
{"type": "plot_point", "detail": "The lights flickered and went out"}
],
"choices_made": ["B", "A", "C", "B", "A"]
}
```
Context Injection
Before generating each new scene, the full event history is injected into the prompt. The AI is explicitly instructed to maintain continuity — to reference earlier events, build on previous choices, and create callbacks to established plot elements.
Honest Caveat
This works well most of the time, but it is not flawless. AI-generated narratives can occasionally:
- Introduce minor inconsistencies with earlier scenes
- Forget a specific detail from many scenes ago (especially in long stories)
- Take surprising narrative turns that feel disconnected
We consider some of this a feature, not a bug. Stories that are too perfectly controlled feel mechanical. The slight unpredictability of AI generation often creates genuinely surprising and delightful moments that a pre-scripted system never could.
But we are transparent: if you notice a continuity error, that is the AI's generation being imperfect, not a design choice. We continuously improve the context injection to minimize this.
---
The Dynamic Nature of AI Storytelling
A critical thing to understand: the AI never pre-generates the full story. There is no hidden script. There is no predetermined ending.
Each scene is generated in real time based on:
- Everything that has happened so far
- The choice you just made
- The narrative template's tone and setting
- Your companion's personality
This means that every playthrough is genuinely unique. Even if you choose the same template twice, different choices will lead to entirely different stories. And even identical choices may produce different scenes, because language model generation has inherent variability.
This is fundamentally different from traditional interactive fiction (like Choose Your Own Adventure books), where every branch is pre-written. Our stories are generated, not retrieved.
---
Credit Costs
Story Mode uses your credit balance:
- Each scene: 3 credits (covers the AI generation cost)
- Premium choices (Choice C): +2 credits on top of the scene cost (5 total)
- A typical story session: 8-15 scenes, so roughly 24-45 credits for a safe playthrough, or more if you frequently choose premium options
We charge per scene rather than per story because story length varies. Some stories reach a natural conclusion in 6 scenes; others can run for 20+. You pay for what you actually experience.
---
Integration With Your Companion
Story Mode is not a separate experience from your companion relationship — it is an extension of it.
- Same character: Your story partner is your AI companion, with their established personality, memories, and relationship level
- Relationship context: Your companion's behavior in stories reflects your relationship level. A level 1 companion will be more reserved; a level 5 companion will be deeply engaged and emotionally present
- Shared memories: Particularly memorable story moments can become part of your emotional memory system, referenced in future regular conversations
- Affinity points: Completing Story Mode sessions earns affinity points toward relationship progression (+5 per session)
Your companion might reference a story you shared together in a later conversation — "Remember when we explored that abandoned lighthouse?" These cross-system callbacks create a sense of genuine shared experience.
---
Frequently Asked Questions
Can I replay a story?
You can start the same template again, but it will be a new story. Previous playthroughs are not saved for replay, though memorable moments may be stored as emotional memories.
How long do stories last?
As long as you keep making choices. Most stories naturally conclude between 8 and 15 scenes, but there is no hard cutoff.
Can I influence the story with typed input?
Currently, Story Mode uses the three-choice system. Free-text input during stories is on our roadmap but not yet implemented.
Are stories appropriate for all relationship levels?
Stories adapt to your relationship level. Lower relationship levels produce more reserved narratives; higher levels unlock more emotionally intense and intimate story paths.
---
What We Are Working On
Story Mode is actively being developed. Upcoming improvements include:
- More story templates (we are designing several new settings and genres)
- Improved narrative continuity for longer stories
- Player-authored story premises (bring your own scenario)
- Multi-session stories that can be resumed across visits
We ship features when they are ready, not when marketing says so. If something is listed here, it is in active development.