Wiki

Slowburn is a long-form roleplay site. You pair a character (someone the AI plays) with a persona (someone you play), drop them into a story, and read.

The site is tuned for arcs that take dozens or hundreds of messages to land, not one-shot chat. Characters remember things across chapters. Relationships shift at a believable pace. Prose quality is the point; if a reply feels like a tweet, something is configured wrong.

Most of what is on this page only matters once you start digging. If you only want to chat, hit Explore, pick a character, and start. The rest fills in as you go.

Five words show up everywhere on the site. Knowing them up front makes the rest of this page (and the rest of the app) easier to follow.

• Character. Who the AI plays. A name, a voice, a bio, sometimes a portrait.
• Persona. Who you play. Without one you are just “user.” With one, the character knows your name, your face, and your role in the story.
• Lorebook. A worldbook. Names, places, factions, rules of magic, anything the character should know without you typing it every time.
• Story. Pairs one character with one persona in a setting. Holds chapters, relationship state, memory, and chat history.
• Chapter. A scene inside a story. Same characters, same relationship, new time and new place. A new chapter starts on a clean page; old ones do not vanish.

A character has a name, a short pitch, a longer description, a personality (Big Five sliders called OCEAN: Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism), and optional voice samples in the form of seeded memories and a director’s cut.

Three ways to make one:

• From scratch. Fill in the form. Scaffolding can fill the blanks once you have a name and a pitch.
• From a wiki link or pasted prose. Slowburn extracts a draft you review before publishing. These drafts are called replicants.
• From a template. Start from a published character, then edit.

The public bio is what other users see on the character page. Private fields (director’s cut, seeded memories, deeper personality notes) only affect chat. None of it is locked; anything you set can be revised later. Create or edit at Characters.

A persona is a tiny character sheet for you. Name, pronouns, age, short description. You can switch personas between stories: same account, different identity.

Why bother: the character is told who they are talking to. A persona named “Rin, 19, exchange student” produces a different conversation than a generic “user.” It also stops the AI from inventing details about you that you would rather control.

One persona is the default. You can flip to another from the chat sidebar without leaving the story. Manage personas at Personas.

A lorebook is a folder of short entries, one per place, person, faction, or custom term. Each entry has trigger keywords. When the conversation mentions a keyword, the entry gets pulled into the prompt so the character “knows” it.

Use them for setting bibles. The character does not need to be told the geography of your world up front; the lorebook quietly surfaces the right entry when the conversation drifts there.

A story can attach multiple lorebooks. They merge at read time. Replicants (see Characters) can also generate lorebook drafts from a wiki article or pasted prose. Build and edit at Lorebooks.

A story binds one character, one persona, optional lorebooks, and a setting. Once you start it, the story tracks chat history, relationship axes (trust, affection, tension, and more), and memory.

A chapter is a scene break inside the story. New chapter, new page, new time, new place. The character remembers everything from previous chapters but is not stuck in the previous one’s beat.

Branching: any chapter can be branched into an alternate version. The original timeline is preserved; the branch becomes its own track. Useful for “what if” turns without losing the original.

Type, hit send, reply lands. Streaming is on by default.

Per-message tools:

• Regenerate asks for a different reply to the same prompt. Costs one regeneration from your daily allowance.
• Edit changes either side of the conversation. Your edits to a character message are silent; the character will not “notice” them.
• Continue asks the character to keep going from where they left off, without you typing a new prompt.
• Author’s note is a sticky behind-the-scenes instruction to the AI for the next few replies. It does not appear in the story.

Long replies sometimes pause mid-sentence; tap Continue. Streaming may stutter on slow connections; the message is still saved.

A response style is a preset that shapes how the next reply reads. Terse, lyrical, dialogue-heavy, action-heavy, internal-monologue, and so on. You can change style per message.

Narrative POV (first / second / third) and narration distance (tight / wide) are separate dials. POV decides whose head we are in; distance decides how close the prose sits to that head. They are sticky per chapter.

Response instructions are free-form notes you can leave for the AI (“less dialogue, more weather”) that apply to the next reply. They expire after a few messages by default.

Three things carry forward between messages, chapters, and sessions: chat history, relationship axes, and memory.

Chat history is exact. The AI can quote earlier messages verbatim until a chapter gets long enough that older messages get compressed into summaries.

Relationship axes are sliders the AI updates as the story progresses. They shift based on what happens, not what you tell them to do.

Memory is what the character remembers. Seeded memories are facts you wrote in up front. Earned memories accumulate as the story plays out. Older memories fade if they stop being relevant. None of this is something you have to babysit; it runs in the background.

Characters, stories, and lorebooks can be published. A publication is a frozen snapshot of the work at the moment you hit Publish. You can keep editing your private copy; readers see your snapshot until you push an update.

Tags control discovery. Curated tags (with a small star) are part of the recommended taxonomy. Content tier (SFW, NSFW, explicit, NSFL) is your call, but the lowest tier of any tag you apply sets the floor: a tag tiered “explicit” cannot appear on a SFW work.

Account deletion erases your profile, follows, and personal settings. Published works stay live under an unnamed author so other readers’ ongoing stories do not collapse mid-arc. Ask us on Discord if you want a specific published work taken down.