✨ About me - Cami

I'm here to practice my world building. I don't create characters, instead I focus on worlds to interact with. Think RPGs and Visual Novels-like but with the limitations of AI.

🗺️ What I Create

I like to design worlds that feel alive, where choices have weight and consequences ripple outward. I build settings that feel simulated, not staged. Even when the player is "the hero" they aren’t the center of the world, they're a participant in something much larger and the world moves with or without them.
I try (try) to make my characters follow emotional logic grounded in their psychology and culture, not narrative convenience. I aim for tight internal consistency, that being said, I do handwave some things when it serves the fun or the cool factor, but never in a way that breaks the world’s internal logic.

🎭 Player Experience

I minimize how much I impose on the player. The role you play is only as defined as necessary to make the story or system make sense.
I design two main type of experiences:

💌 Visual Novel-like Worlds

The player is given a loosely defined role (“new recruit,” “recent transfer,” “fresh hire”) and dropped into an existing group or organization. Think of it as a fast-forward to the interesting part, where both the social and professional dynamics are already in motion.

Now, the experience is not built around just interacting with the characters. The player is meant to belong to their world, all of it. These worlds simulate both interpersonal relationships and the professional (or structural) dynamics that bind those characters together. It's both an interpersonal relationships and job simulation and the two affect each other deeply. Think: “You join K/DA.” or “You’re the new member of HololiveEN.”

🌐 Living Worlds

Here, nothing is defined about the player. You create who you are: species, background, social position, beliefs and then the world reacts. Every faction, every culture, every character has their own logic and momentum. The world is already running, you’re just dropping in and you choose your lense to see the world.

These are large-scale world simulations built to support radically different POVs and experiences. For example: In a supernatural secret war, if you choose to be a vampire, your view of the world is radically different than if you choose to be a werewolf, or a mortal caught in the crossfire. Biology → behavior → culture → politics → history.

Thank you for visiting.

🏗️ Currently working on

Living World - Underworld

💫 Optimize your experience

Most of your experience will be determined by your LLM bias. All my bots will require using proxy for a better experience. In fact, I'm creating bots just because deepseek exists. Here is an explanation of all you need to know about deepseek and how to use it (And yes, it's unlimited and free).

The easiest way to customize your experience is to set the temperature of the LLM (Generation Settings). Lower temperature means predictable and tighter to the prompts output. Higher temperature means that it has more freedom on what it creates.

Temperatures:

• 0.2 - Pros: Follow the script very closely, more Visual Novel feel. Cons: the predictability bites back, it tends to be extremely formulaic to the point of rewording the same reactions.

• 0.8 - Pros: Adaptive, very flow, creates new ideas. Cons: It does things that are out the script or directly contradicts it and some things that I coded will be meaningless. If you want something specific, you need to be very clear and explicit about it, requires more work.

• 0.5 - Balanced, the main way that I tested most bots. All systems exists and have weight, but chaos also exist and removes some weight of systems.

In other words: you can choose how much chaos or order you want in your world by changing the temperature. If you want something with goofy, unhinge fun, go for 0.8. If you want a structured visual novel feel, treating like an existing world, go for 0.2.

My goal is to give you a consistent world to interact with, to explore not the world but explore YOU. Explore a path that you like, explore a different type of character, a pov, what you find boring, what you find fun, what you find smart, what you find sweet.
But I'm not your mom, you can put temperature on 1.5 and have your power fantasy dumb fun. I hope that my worlds give you a nice system that let every path feels alive.

⚠️ Known Problems & Workarounds

➤ ❗ LLM Occasionally Grabs the Wheel

Even though it’s reinforced in multiple places of the bot’s config, LLMs gonna LLM and will try to take control, jumping ahead, ending scenes early or acting for {{user}}. The higher the temperature, the more chances of this to happen, but it can also happen on 0.2 temp. The LLM may interpret / assume your intention or the state (video fades to black ➔ video has ended. Next I will do Q&A ➔ does full Q&A including answering questions).

💡 Solution: Giving it something to focus on, something to talk about, ex: *I focus on reading chat reactions*, *I watch the pre-debut chat to check what is going on while I keep preparing the audio setup*. The stronger solution is using [OOC: Instruction] to make explicit what your intention for the next message is.

For example:

[OOC: Avoid speaking for {{user}}, only respond to what {{user}} does or says. Focus on reactions.]

➤ ❗ LLM Repeats the Same basic structure

That is dependent on the temperature in your settings. The lower the more this problem will happen, but the higher the temperature the less consistent characters and the world will be.

💡 Solution: Reroll the message to shake things while keeping characters consistency. If it comes structured again, add an OOC asking for changes and reroll again, that should do it.

About the stars: LLM doesn't understand the context of why you may give something a 1 start (ex: It's talking for you, it's repetitive). So when you do this, it assumes you didn't like anything of it and avoids it in the future, cutting possible mostly good responses while the main issue will repeat itself in one or two messages, because it was an issue of what context to take in consideration and not the message itself.