Creating with chat

Chat is where everything starts in MotherFluxer. You describe what you want, and MotherFluxer makes it. The same chat box can produce any kind of artifact, so you never have to pick a tool or a model up front.

Write your first prompt

Find the chat box marked "Ask anything…" and tell MotherFluxer what you want to create. Be specific and lead with the thing itself.

Strong prompts name:

• What it is — an image, a video, a song, a 3D model, a document, or an App.
• The subject — what's in it or what it's about.
• The style or detail — mood, color, length, format, or anything that matters to you.

Examples:

• "A watercolor poster of a fox reading under a streetlamp."
• "A 30-second upbeat synthwave track for a workout video."
• "A 3D model of a low-poly spaceship."
• "A one-page project brief for a community garden, with goals and a timeline."
• "An App that's a tip calculator with a slider for the percentage."

Press Enter to send. You can write in plain language — no special syntax, no keywords.

One chat, any artifact

A single chat can create anything. You don't switch surfaces to go from an image to a song to a document — you just ask. MotherFluxer reads your request and builds the right type of artifact for it.

• Image — pictures, posters, illustrations, logos.
• Video — clips and short scenes.
• Audio — music and sound.
• 3D — models you can rotate and view.
• Document — formatted text, briefs, outlines, write-ups.
• App — small interactive tools that run right in your workspace.

If your prompt is ambiguous, say which type you want ("make this a video, not an image") and MotherFluxer will follow your lead.

Attach a file to your message

The attach button at the left end of the chat box adds a file to the message you are writing — a photo to work from, a document for the AI to read, a video for it to watch. Each one shows as a small tile above the box with its own remove button, and one message holds up to five. Removing a tile before you send also deletes that upload.

• A send waits until every attachment is ready. The send button holds back while a file is still uploading. A photo whose preview is still being made after its upload — an iPhone photo added in a browser that can't open it, for example — is waited for, and if it isn't ready within about a minute the message stays unsent with An attached picture is still being prepared, so the message was not sent. Try again in a moment.
• A video may be reformatted first. When the AI that will watch your video can't take the file as it is, your device makes a copy it can take before the message goes, and the video's tile reads Reformatting video at model's request with a ring showing how far it has got. If the copy can't be made, the message is held back with An attached video could not be reformatted for the AI, so the message was not sent. Try again. — the AI never answers as if the video weren't there.

Build a 3D model from a picture

You can start a 3D model from a picture instead of from words alone.

• Attach the picture to your message with the attach button at the left end of the chat box, then ask for a 3D model of it — for example, "Make a 3D model of this lamp."
• MotherFluxer answers with a suggestion for a new 3D model, showing its estimated cost and the description it will use. Select Generate to build it.
• The model is shaped from your picture, and your words only refine it. Building it uses credits, like any 3D model.

A 3D model is the one new artifact that can be built from a picture in chat. When you want to change the picture itself, describe the change instead and MotherFluxer suggests an edit to it.

You don't pick the AI — MotherFluxer routes for you

There's no model menu to wrestle with before you create. MotherFluxer reads your request and automatically routes it to the right models for the job, so a music prompt goes to what's best for music and an image prompt goes to what's best for images.

When a step does offer a choice — for example, different quality or speed options — you'll see those options laid out with each one's cost and speed, and you simply pick the one you want. Otherwise, just describe what you want and let MotherFluxer handle the rest.

Watch responses stream in

After you send a prompt, the response streams in as it's generated, so you can read along in real time instead of staring at a blank screen.

When a piece of work finishes, it appears right inside the chat as a card — a compact preview of the artifact you just made. Text answers stay in the chat as text; finished images, videos, audio, 3D models, documents, and Apps show up as cards you can act on.

Every card rests as a picture of what you made, so a long chat stays quick. Tap a card to open what you made in its editor — except an App, which a tap runs right in its card. Any one card can come to life at a time, and starting one stops the one before:

• App — click or tap it and the App starts inside the card, and you use it right there: press its buttons, type into it, scroll it. It arranges itself the way it would in a window the card's size. While it starts, the card says so — an App with its own server can take up to a minute to wake. On a computer, resting the pointer on some Apps shows a preview first; the preview doesn't take presses until you click. To work on the App instead, choose Open in editor from the card's menu. An App with nothing to show in a window — only code, or only a server — says This App doesn’t run here and points you to the editor.
• 3D model — scroll it to the middle of the chat and it turns on its own; drag to turn it yourself, and pinch or scroll to zoom.
• Audio and video — press the play button on the card. The time beside it shows how far it has played and how long it runs; drag along the card to scrub.
• Image — on a chat someone shared with you, tap the picture to see it full size. Pinch, scroll or double-tap to zoom in, drag to look around, and press Escape or tap beside the picture to close it.

When you press Generate

Sometimes MotherFluxer answers with a suggestion instead of making something straight away — a new artifact, or a change to one you attached — showing what it will do and its estimated cost. Select Generate to go ahead.

• The result lands in the same chat. When the work finishes, its card appears beneath the suggestion, and the chat stays where you are — you are never taken out of the conversation. An App the response already shows — one MotherFluxer started while it answered — is finished on that same card rather than in a new one.
• Generate becomes Open. Once the card has landed, the suggestion's button opens that result, so you can jump straight to it later.
• A change that needs your input opens the editor. If the suggestion needs something only you can give — an area to draw, a picture to pick — its button reads Set up & generate and opens the editor with the change set up, so you finish it there.
• A suggestion of several changes runs them in order. Its button says how many steps it will run, such as "Generate 3 steps". If one step fails, the changes the earlier steps made are kept.

While MotherFluxer works

A short line beside the spinning icon says what is happening — reading your message, looking at the files you attached, searching the web — and changes as the work moves along.

• Longer jobs show how long they have been running. A video, a 3D model or a song can take a few minutes, so the line underneath counts up as it goes and, once MotherFluxer has made enough of that kind of thing, tells you how long it usually takes.
• If the AI service is busy, you'll see it retrying. MotherFluxer tries again on its own; you only need to act if it says it couldn't finish, and then a Try again button appears when another attempt could work.
• Under an answer, quick buttons such as Tell me more, Explain simply and Give examples send that request for you.

All of these lines and buttons appear in the language you chose in Settings.

Knowing you're talking with an AI

A short line above a reply reminds you that the answers come from an AI: "You're chatting with an AI, not a person."

• It opens every chat. The line sits above the first reply.
• It comes back during a long chat. Any reply that arrives three hours or more after the line last appeared carries it again.
• It reminds you to take a break. When that reply continues a chat you've kept going, the line also says you've been chatting for a while and to remember to take a break.

The lines stay where they appeared, so a chat you come back to later shows them in the same places.

Getting help in a crisis

If a message you send suggests you may be thinking about suicide or harming yourself, MotherFluxer answers with support, and a card beneath the reply shows where to reach help right now:

• In the US, call or text 988 to reach the Suicide & Crisis Lifeline — free, confidential, any time. Call 988 and Text 988 on the card start the call or the message.
• Outside the US, findahelpline.com lists a free helpline in your country.
• If you are in immediate danger, call your local emergency number.

The numbers and links on the card are part of MotherFluxer itself — they are never written by the AI. This applies to everyone who uses chat. How MotherFluxer recognizes these messages and responds to them is published on the Safety page.

Stop a response part-way through

While a response is streaming, the send button becomes Stop. Press it and the work stops right there, wherever it had got to.

• You keep what you already saw. The part of the reply that had arrived stays in the chat as a normal message, so you can read it, copy from it, or reply to it. It is marked as stopped rather than as an error — you won't get an error message or be asked to report a problem, because nothing went wrong.
• You're charged for the work that had already been done, and nothing more. If MotherFluxer had already handed a generation off to be produced, that piece is paid for whether you wait for it or not; anything it hadn't started yet costs you nothing.
• Stopping doesn't end the chat. Send your next message straight away — a stopped turn leaves the chat in a normal, ready state.

Stopping is the right move when you can already tell a reply is going the wrong way. Rather than waiting for it to finish, stop it and rephrase.

Sending the same thing twice

If you press send twice in quick succession — or a flaky connection makes your device retry on its own — MotherFluxer runs it once, and you are not charged twice. In chat the duplicate is simply folded into the request already running, and the one answer appears as normal. Elsewhere in the app, in Smart Edit or when you're chatting with support, you'll see "Already sent" instead, telling you the first attempt is still on its way.

A deliberate second press after the first one has finished is a different matter: that's a new request, and it runs and is charged as one.

Speak your message instead of typing it

You can dictate a message straight into the chat box. Your speech is turned into text on your own device — the audio is never sent anywhere, and nothing is downloaded to make it work.

• Look for the dictate button at the right end of the chat box, where the send button goes. It stands there while the box is empty and gives way to send once you type; if an empty box doesn't show it, your browser or phone can't do this.
• Tap or click it and start talking. The words appear in the chat box as you speak. Typing in the box stops dictation and keeps every word already there.
• Tap the same button again to stop. The text stays in the box, so you can edit it before you send.

Dictation uses your one language setting — the same setting that controls the app's wording. There's no separate speech language to pick.

What to expect on a phone versus a browser

On an iPhone or an Android phone, your phone's own dictation adds punctuation and capital letters for you, exactly as it does everywhere else on the device.

In a web browser it does not. No browser adds punctuation or capitalization to dictated text in any mode, so you'll get a run of unpunctuated words. You can speak the punctuation you want — say "comma", "period", or "question mark" — or add it yourself before sending. This is a limitation of browsers, not something a setting can turn on.

When the dictate button isn't there

The button only appears when your device can actually do it, so you're never offered something that would fail:

• Firefox has no speech recognition at all, so dictation isn't available there. Chrome, Edge, and Safari all work.
• On a phone, dictation needs a recent operating system — iOS 26 or newer, or Android 13 or newer — and your phone must already have your language installed for its own dictation. If it doesn't, the button stays hidden.

Have a response read aloud

Any text response in chat can be read out loud in a natural, expressive voice, so you can listen instead of read — handy when you're multitasking or just want to hear how something sounds.

• Look for the read-aloud button in the row of actions beneath an assistant response.
• Tap or click it to start listening; the response is spoken from the beginning, and the button shows it's playing.
• Tap the same button again to stop. Only one response plays at a time, so starting a new one stops the last.

Playback begins within a moment of tapping and continues straight through to the end, even for a long response. The voice speaks the language you chose in Settings — English or Spanish — so a response written in another language is read with that voice's accent.

Copy or share a response

The row of actions beneath a response can also hold copy and share.

• Copy puts the response's words on your clipboard, with a short "Copied" note. If the response cites sources, each one shows as a number like [1], with the sources listed underneath.
• When the response made a picture, copy puts the picture on your clipboard too, so you can paste it into another app. Videos, sounds, 3D models, Apps and documents can't go on a clipboard, so copy takes the words only.
• Share opens your device's own share menu, the same one other apps use. It sends the file the response made, like the picture, video or sound, along with any words. When the menu can't take that kind of file, it sends the words.
• If the file is still on its way when you press share, a note says "Ready to share. Tap share again." — press share once more and the menu opens.
• A browser with no share menu of its own offers copy only.
• Which apps appear in that menu, such as WhatsApp, Messages or Mail, depends on your device and the apps installed on it.
• Sharing never sends a link to your chat, so nothing you didn't choose to send leaves your account.

Save a response as a document

Some responses also offer save as document in the row of actions beneath them. Pressing it keeps the response's words as a new document in your gallery, so you can open it, edit it and share it like anything else you've made.

• The document holds the same words copy would — lists, tables and code as written, and any sources listed underneath.
• It is named after the chat it came from, and is marked as AI-made, like every file the AI writes.
• A note says "Saved to your gallery as a document." when it's done. The chat itself doesn't change.

Open an artifact from chat

Each finished card is a doorway into the full artifact.

• Tap or click the card to open the artifact in its editor. On a chat someone shared with you, tapping an image's picture opens it full size instead. A drag on a 3D model turns it rather than opening it.
• An App runs when you tap it, so open it with Open in editor from the card's menu.
• From there you can play it, view it, adjust it, or fine-tune it with the editing tools for that type.
• Close the editor to return to your chat right where you left off.
• A card that says Deleted, under its type's faded icon, is an artifact that was deleted — from Artifacts, its editor or the card's own menu. Something you attached to a message and later deleted shows the same card. It has no menu and can't be opened, and the rest of the chat stays as it was.
• A card that says Couldn't make this, under the same faded icon, is something that could not be made — the AI couldn't finish it, so there is nothing to open. Every kind of artifact shows the same card, and it changes to it by itself once the making gives up; ask again in the chat to try once more.

Your artifacts are saved automatically, so you can come back and reopen them later from Artifacts.

Keep the chat going to refine

You don't have to nail it on the first try. Chat remembers what you just made, so you can refine by replying in plain language.

• "Make the background darker."
• "Add a second verse."
• "Shorten it to one paragraph."
• "Try it in a brighter color palette."

Each reply builds on the work already in the chat, so refining feels like a back-and-forth rather than starting over. Keep nudging until it's right.

Start a new chat

When you're moving on to something unrelated, start a fresh chat so the new one isn't shaped by the last one.

• Use the option to start a new chat to get a clean "Ask anything…" box.
• A new chat means a clean slate — MotherFluxer won't assume your next request relates to the previous artifacts.
• Your earlier chats and their artifacts stay saved, so nothing is lost when you begin again.

Reach for a new chat whenever you switch topics; keep replying in the same chat when you're still refining the same idea.

Find an earlier chat

Your recent chats are listed in the sidebar. When the one you want isn't there, click All Chats… at the bottom of that list to open the full Chats screen, which holds everything you've ever started.

• Search across everything. The search box looks through chats, messages, Projects, and artifacts at once, so you can find a chat by something that was said inside it, not just by its title.
• Narrow by what you're looking for. Four tabs — Chats, Messages, Projects, Artifacts — each show how many matches they hold, so you can jump straight to the kind of result you want.
• Change the order. Toggle between newest-first and oldest-first.
• Keep scrolling. Long lists load in pages; use Load more to pull in the rest.

Click any result to reopen it exactly where you left off. Use Exit All Chats to go back.

Rename, favorite, or delete a chat

Every chat has a menu — on its row in the sidebar, and on its row in the Chats screen — with three actions.

• Rename — chats are named automatically from your first prompt. Rename one to something you'll recognize later; the new title saves as soon as you confirm it.
• Favorite — moves a chat to the top of the sidebar list and marks it with a star, so the ones you keep returning to stay in reach. Choose Unfavorite on the same menu to undo it.
• Delete Chat — removes the chat and its messages. You'll be asked to confirm, because this cannot be undone.

Deleting a chat does not delete the artifacts it produced — those stay in your Artifacts. If a chat belongs to a Project, removing it from the Project is a separate, non-destructive action that leaves the chat itself intact.

Use a ghost chat for a one-off

A ghost chat is a temporary, throwaway chat. Use it for quick, one-off generations you don't need to keep around. When you navigate away, the ghost chat is automatically deleted and never shows up in your chat history.

Reach for one when the chat itself doesn't matter, only the result does:

• Quick experiments and "let me just try this" prompts.
• One-off generations you'll grab and move on from.
• Throwaway ideas you don't want cluttering your saved history.
• Sensitive or personal prompts you'd rather not leave behind.

If you're starting real, ongoing work you'll want to return to, use a normal chat instead.

How a ghost chat differs from a normal chat

A normal chat is saved automatically. It stays in your history, you can reopen it later, and it can live inside your Projects alongside the rest of your work.

A ghost chat behaves the opposite way:

• It isn't saved. The chat is temporary by design.
• It disappears when you leave. Navigate away and the ghost chat is auto-deleted.
• It stays out of your history. You won't find it later, because it was never kept.

Everything else feels the same. You chat, generate, and edit artifacts exactly as you normally would, including Smart Edit, Direct Edit, and the File menu.

A ghost chat isn't kept, but your artifact can be

The ghost part applies to the chat, not to what you create.

• The chat is gone for good. Once you navigate away, the messages can't be recovered.
• An artifact you produced can still be saved. Save it before you leave and it lives on in your Artifacts like any other creation, ready to edit, export, or publish.

So if a ghost chat produces something worth keeping, save the artifact first. Then leave it behind, knowing the chat won't be stored but your work will.

A generation didn't finish or errored

Generations occasionally fail or stall. When that happens:

• Give it a moment first. When the model handling your request stops responding, MotherFluxer quietly hands the request to a different model and answers from that one instead. You may notice a reply arriving a little slower than usual; you should not see an error. This happens once — if the second model also fails, you get an error you can retry.
• Try again. Re-run the same request — many errors are transient and clear on a second attempt.
• Rephrase your prompt. Make it more specific, shorter, or clearer about the result you want.
• Pick a different model option. If the request lets you choose a model, try another one from the options shown — each has its own cost and speed.

If it keeps failing, simplify the request to its core idea, then add detail back once you get a result.