TL;DR
- What it is: A character-based conversational platform where you can create or pick AI personas and roleplay immersive scenes—or just chat.
- How it works: Characters run on large language models (LLMs). You can connect your own API key (e.g., OpenAI/Claude) or use JanitorLLM (when available).
- Who it’s for: Adults (18+) who want creative, personality-driven conversations, writing help, or roleplay.
- Status note: The team has communicated a UK access block related to regulatory requirements.
What Exactly Is Janitor.AI?
Janitor.AI (also written Janitor AI or JanitorAI) is an online hub to build, share, and chat with AI characters. Browse thousands of community-made personas—from fantasy heroes to everyday coworkers—or craft your own with custom prompts, greetings, and example dialogues. It’s community-driven and optimized for free-form storytelling as well as casual chat.
Core use cases
- Creative writing & worldbuilding: Draft scenes, test character chemistry, and iterate story arcs in dialogue.
- Roleplay & improvisation: Spin up dynamic scenarios with clear rules, goals, and constraints.
- Practice & play: Language practice, social rehearsal, or just decompressing with a character that follows your brief.
Pro tip: Give your scene a short “director’s note” (role, tone, goal, boundaries) before the first message. Coherence jumps noticeably.
How Janitor.AI Works (Plain English)
- Pick or create a character. Each character has traits, a greeting, and sample dialogue that shape its voice.
- Choose a model.
- Bring your own API key (e.g., OpenAI, Anthropic/Claude) for speed and quality; usage is billed by the provider.
- Use JanitorLLM (free/beta when available) if you don’t want to connect an external key.
- Chat. The platform tracks context within the session and stays in character. You can regenerate or edit messages.
- Tweak & iterate. Adjust persona, tone, memory notes, or start a fresh session when the thread drifts.
Features That Make Janitor.AI Stand Out
- Seamless chat ↔ roleplay: Switch from normal chat to in-character storytelling without changing tools.
- Deep persona control: Define tone, boundaries, backstory, and example messages; save reusable prompts.
- Community character library: Discover trending bots, follow creators, remix ideas, and tag characters.
- Flexible model routing: Swap between your API key and JanitorLLM to balance speed, quality, and cost.
- Session context: Characters reference earlier messages in the same chat to keep stories coherent.
- Quality controls: Regenerate, edit, or constrain style (e.g., “3–5 sentences,” “end with a question,” “avoid purple prose”).
Getting Started: Step-by-Step
- Create an account and pick a starter character.
- (Optional) Add an API key
- Generate a key from your provider’s dashboard and paste it into Janitor.AI’s API settings.
- Open a chat and set a scene with a short brief:
- Role: “You’re Captain Elara, a pragmatic starship engineer.”
- Tone: “Calm, dry humor.”
- Goal: “Diagnose a power fault before arrival.”
- Constraints: “No technobabble; ask questions first.”
- Iterate: If replies drift, restate constraints or reset the session.
- Save your best prompts for reuse across characters.
Pricing & Access
- Platform access: Frequently free, especially with JanitorLLM when available.
- Model costs: If you connect OpenAI/Claude, their metered pricing applies.
- Reality check: Community reverse proxies can be slow or unstable; a personal API key is usually faster and more reliable.
Safety, Privacy, and Content Policies
- Adults only (18+).
- NSFW allowance within strict boundaries: Adult roleplay is permitted, but prohibited themes include minors and other disallowed content. Always review the latest rules before engaging.
- Privacy basics: Avoid sharing personal, financial, or medical info in any AI chat. Use neutral usernames, and save only what you’re comfortable retaining.
- Control the experience: Set clear boundaries in your brief; stop or reset if a thread heads somewhere you don’t want.
Janitor.AI vs. Traditional Chatbots
Use case | Traditional bots | Janitor.AI |
---|---|---|
FAQs / task completion | Excellent (structured flows) | Possible, but overkill |
Long, creative dialogue | Limited | Excels |
Persona / roleplay fidelity | Minimal | Deep persona controls |
Cost predictability | Often fixed plans | Varies by API usage |
Prompts & Patterns That Work
- Director’s note before first message (role, tone, goal, boundaries).
- Memory snippets: One-liners the character should remember this session.
- Turn discipline: Ask the character to end every reply with a question that moves the scene forward.
- Style gates: “Keep answers ≤ 8 sentences,” “Use plain style,” or “Ask 2 clarifying questions first.”
Reusable starter prompt (copy/paste):
“You are {{role}}. Tone: {{tone}}. Goal: {{goal}}. Constraints: {{constraints}}. Stay in character. Keep replies under 8 sentences and end with a question that advances the scene.”
Troubleshooting Quality
- Too flowery? Add: “Use plain, concise style; avoid purple prose.”
- Losing plot? Restate the goal and constraints; restart the scene if needed.
- Repetition/rambling? Ask for bullet points or numbered steps.
- Model hiccups? Switch models (your API key vs. JanitorLLM) or regenerate the last response.
FAQs About Janitor.AI
Is Janitor.AI free?
Using JanitorLLM is typically free when available. If you connect OpenAI/Claude, your provider’s metered pricing applies.
Does Janitor.AI allow NSFW content?
Yes—within strict boundaries. Absolutely no minors or other disallowed themes. Review current rules before engaging.
Do I need an app?
No. Janitor.AI is web-based and works on desktop and mobile browsers.
What is JanitorLLM?
A free/beta in-house model that some users can access. Performance and availability can vary.
Why can’t I access from the UK?
The service has communicated a UK access block related to regulatory requirements. Users in the UK may see an access-blocked page.