KOORI Journal

Friendly conversations

AI Characters You Can Just Talk To

KOORI Journal ·

Sometimes you do not want a quest, a dramatic confession or a complicated character biography. You want to talk about the book on your table, a meal you might cook or a small idea you have not quite finished. Friendly AI chat can be a light creative space for that kind of conversation, provided you keep clear what it is: an interaction with a fictional AI character, not a human relationship.

A friendly green fictional creature reading in a lamplit nook beside a cat and rainy window
Original AI-generated editorial illustration commissioned by KOORI. Illustrative, not an app screenshot.

Find a conversational fit, not a perfect portrait

A striking image can get your attention, but the character description gives you a better clue about the conversation. Look for curiosity, a gentle sense of humor or a concrete everyday interest. A librarian with opinions about messy shelves offers more to talk about than a companion who promises to be perfect at everything. Fictional preferences create room for a back-and-forth.

In Discover, browse the available characters and style or World options, then open a character to learn more. Choose according to the mood you want today rather than treating the choice as permanent. A small monster can have a thoughtful voice; an elegant fantasy character can enjoy ordinary conversation. Artwork style does not reliably tell you whether the chat will be romantic, adventurous or calm.

Set a simple expectation early: friendly and platonic, no flirting, short replies, one question at a time. This is a message you can write, not a promised relationship-mode switch. AI can misread instructions or drift. If the response feels wrong, correct the tone, choose another character or stop. You do not owe a fictional character your time, attention or personal disclosure.

A step-by-step first friendly conversation

The current beta requires an account before you can try the app. Use the available sign-in provider, complete eligibility and onboarding, then return to Discover. This guide does not promise unlimited access to every creative or calling feature. Check the app’s current credit and plan information before starting a paid action; ordinary text conversation and generated media are different experiences.

Start with one low-pressure topic

  1. Step 1: Open Discover and choose a mood. Browse the character descriptions and available filters. Pick a character with an interest you can respond to, such as books, cooking, plants or curious inventions. You can change your mind later. Avoid assuming that a cute portrait guarantees a particular personality or a call-ready avatar.
  2. Step 2: Open the character and enter chat. Use the character’s chat action. Read the opening message and notice one detail you might follow. If the opener is more dramatic than you want, you can set a calmer scene in your own first message rather than continuing a premise that does not suit you.
  3. Step 3: State the tone directly. Try: “Let’s have a friendly, completely platonic conversation. Keep replies brief, do not flirt, and ask one question at a time.” You can request dialogue without long narration. Describe what you do want as well as what you do not want, so the AI has a positive direction.
  4. Step 4: Offer one ordinary starting point. Mention a fictional or nonsensitive topic: choosing a rainy-day book, inventing a soup recipe or deciding what a tiny space garden should grow. Ask a question with enough detail to answer. “Which of these two books fits a quiet evening?” gives more direction than “Entertain me.”
  5. Step 5: Follow a detail instead of restarting. If the character suggests a mystery, ask what makes it cozy. If it invents a soup, ask for a silly name. One follow-up can turn a generic response into a shared thread. Correct factual claims you intend to act on, and do not treat character confidence as evidence of accuracy.
  6. Step 6: Pause when you want to. You can end the conversation without an explanation. If you return, a short recap of the topic is useful. Do not assume flawless permanent memory. Keep real-world routines and human relationships separate from this optional fictional activity, and use the privacy policy to understand how the service handles data.

Everyday starters that leave room for a reply

The best starter asks for one manageable contribution. “I am inventing a cafe menu for a rainy afternoon. What should its first drink be?” invites an opinion and a follow-up. “Tell me everything about yourself” often produces a biography instead of a conversation. You can be specific without being personal: fictional errands, imagined gardens and book preferences are useful material.

For playful conversation, introduce a small constraint. Design a sandwich that a dragon could eat without burning it, choose a name for a robot’s houseplant or plan a library where the shelves move on their own. Constraints give the character something to solve and let you disagree gently. If every idea is accepted immediately, ask for a preference and a reason.

For a quieter exchange, ask for less. One observation and one question may be enough. Try a discussion of textures, music moods or an imaginary walk rather than an emotionally demanding scenario. You can request that the character avoid giving advice and simply explore the topic. There is no requirement to turn an ordinary conversation into a roleplay plot or a personal confession.

Let’s keep this friendly and completely platonic, with short replies and no long scene narration. We are inventing a cozy reading cafe. I think every table should have a different lamp. What small detail would you add, and why? Ask me just one question afterward.

Set boundaries without building a complicated persona

You do not need to write a long specification to ask for a respectful tone. State a handful of preferences: no flirting, no pet names, no assumptions about your feelings, no unsolicited advice. If a response violates one, correct the specific behavior rather than rewriting the entire conversation. For example: “Please use my chosen name, keep this platonic and return to the cafe idea.”

Optional scene description can be enjoyable, but it is not required for friendly chat. If the AI narrates your actions, tell it to leave your actions to you. If the reply is too long, ask for a few sentences. More elaborate prose is not inherently more caring or more useful. A readable exchange on a phone often works better when it leaves you an easy next turn.

A character may have a strong fictional personality that does not match your request. You can try a different one rather than trying to make every character behave identically. If you want an original gentle companion, the creature guide shows how to describe a personality through habits and small choices. That is a creative design exercise, not a guarantee of a reliable emotional support system.

Privacy, memory and the limits worth knowing

Use fictional names and avoid passwords, addresses, financial information, private medical details or other people’s confidential messages. Do not assume a chat is secret simply because it looks like a personal conversation. Read KOORI’s privacy policy for how data is handled, and use available account controls. This guide does not promise automatic deletion or confidentiality from every service involved.

AI can sound certain while inventing a fact, misunderstand a boundary or lose track of a previous detail. Conversation history and memory-related features do not establish perfect permanent recall. Remind the character of the relevant context when needed. For factual questions, consult reliable sources separately, particularly before making a health, legal or financial decision.

An AI character is not a person, therapist or emergency service. Do not use it as your sole support for distress or a substitute for professional care. If you need urgent help, contact an appropriate local service or a trusted person. Friendly fictional conversation can be a creative pastime without making promises about love, understanding, consciousness or human-level care.

Troubleshooting and friendly-chat FAQ

If the conversation keeps becoming romantic, clearly restate the platonic boundary and choose another character if the tone still does not fit. If replies repeat, introduce one new object or ask the character to compare two options. If it forgets the cafe plan, summarize the last decision. If a message fails to send, check your connection before repeatedly sending the same text. An imperfect exchange is a reason to adjust or pause, not to disclose more personal information.

Do I need to create a character first?

No. You can sign in, complete onboarding and choose an available character from Discover. Creating your own is an optional path when you want more control over the fictional personality and visual direction.

Is there a guaranteed platonic mode?

This guide sets the tone through clear character and chat instructions. It does not promise a dedicated relationship toggle or flawless compliance. Correct drift, switch characters or stop if the conversation is uncomfortable.

Will the character remember everything?

Do not assume perfect or permanent recall. AI may lose context or confuse details. A concise recap helps when continuing a topic, and a separate note is useful for facts you want to preserve.

Can I call any character?

No. Live video calls depend on the character’s completed avatar setup, availability and your call budget. Phone calls are currently coming soon. Text chat is the simpler choice when you only want an ordinary conversation.

More from KOORI Journal