Witbitz docs HomeTrustAll docs

Witbitz Spaces — user guide

What Spaces is. A Space is a shared chat room with an AI assistant in it. You can keep a Space to yourself, or invite people and talk with them and the assistant together. A Space's messages are stored encrypted with its own key, which travels in its link. A regular Space opens for anyone with its link; a verified Space lets only the people you list post, and they prove who they are with Google or an emailed code; a confidential Space runs its assistant in an attested secure enclave, so not even Witbitz can read it. You can also make temporary Spaces that are deleted when unused, and direct chats with one other person.

Installing

Spaces installs from the browser like an app — no app store and nothing to sign up for. Open the install page of the app your Spaces are in, on the device you want it on; the page shows the steps for that device.

Your Spaces open only in the app they were made in. Witbitz Spaces (witbitz-spaces.pages.dev), the hosted app (app.witbitz.chat) and apps others build on Witbitz are separate apps, each with its own install page (its address + /install, e.g. witbitz-spaces.pages.dev/install). A new computer or phone must install the SAME app as your other devices: "Add a device" refuses a code from another app ("That device is setting up a different app…") before anything is sent. Each app also keeps its own account: your connections, Code computers and Code's monthly allowance, Siri and outside agents belong to the app you set them up in — the same email in the other app starts empty.

What "installed" means: there is no app store — Spaces is installed by the browser, and an install belongs to that browser and its profile. Installed from Chrome is not installed in Edge, and a second Chrome profile (another person or account in the browser) installs separately. On a computer and on Android, the installed app and the browser tabs of the same browser and profile share their Spaces; a different browser or profile starts empty until you restore or bring your Spaces over. On an iPhone, Safari and the Home Screen app are separate even on the same phone. Already installed? Open it from the Start menu, Dock or home screen, or with "Open in app" in the address bar. The install page notices when Spaces is already set up in that browser (signed in, your Spaces there) and offers "Open Spaces" instead of the install steps.

In a browser tab Spaces opens only a link to a Space someone sent you. Everything else — creating your account, restoring it, the list of your Spaces, New Space, Code — opens only in the installed app; in a browser tab the page goes to the install steps instead. (On an iPhone a browser tab and the installed app keep separate storage, so an account made in the tab would never reach the app.)

On a computer (Windows, Mac, Linux, ChromeOS) the first step is "Secure this computer" — tap "Secure with a passkey": one passkey (Windows Hello, Touch ID) encrypts everything Spaces keeps in this browser, and the same passkey becomes one of your account's backup passkeys, so setup asks once. If this computer's passkey can't hold the key (some password managers), "Use my phone instead" saves it on your phone, and every unlock then asks for the phone. A computer that already has Spaces asks for this the next time it opens. A Space link still opens before you do it, but nothing of it is saved on the computer until you choose Secure. Phones and tablets skip this step.

After installing: open the app. "Create your account" needs an email address — you sign in with a 6-digit code sent to it (or with Google) — and then gives you a recovery code so you can get your Spaces back on another device. If no account is listed for that email, it asks first — "Create a new account?" — because an account can be unlisted: if you already have one, "Restore instead" brings it back with your recovery code or passkey; a new account starts empty. (If it can't check just now, it says "Couldn’t check for your account" and asks the same.) Next it offers a passkey (Face ID / Touch ID instead of typing a code) — optional: "Not now" skips it, and you can add one any time in Settings → Privacy & security → "Back up & recovery" → "Add a passkey" (on a computer, the passkey that secured it is used). Or tap "Restore your Spaces" to bring them back with your passkey, your recovery code, or another device that has them (see Devices, sign-in and recovery). It offers notifications when you first open it — say yes, or you won't know when someone replies.

Install help: "Need help installing? Ask" on the install page, "Ask for help" in the Restore sheet of a new install, and "Need help?" at the top of the screen for the whole of setting up — the welcome screen, signing in, the passkey and the recovery code, until the app lets you in — open a conversation with the Help assistant about installing and getting started, as a card: from the bottom on a phone, at the side on a computer, with the screen you were on left where it was. No sign-in is needed. It is a temporary confidential Space: not listed among your Spaces, kept on this device for 7 days after your last question, then deleted — and nothing written to it is kept more than 8 days. It can read this guide and see which screen you came from and what you are using — phone, tablet or computer, its system, the browser, and whether the app is installed — so its steps match your device; nothing else. It allows 20 questions a day from each network; with Hide my IP on, everyone using it counts as one network. On an iPhone, Safari and the installed app each have their own conversation.

Not available when: today's questions are used up — it says "Install help has answered as many questions as it can today. Try again tomorrow." The steps above still apply.

Help

Help is your own assistant for using the app. Ask it how to do something, why something isn't available, about your own Spaces, settings and devices ("how many Spaces do I have?", "is the lock on?", "which devices are signed in?"), where something was said ("where did we decide the hotel?"), or where a photo or file was shared ("the PDF about the lease", "the photo of the beach") — files by name and type, and a photo by what it shows when the Space's assistant has looked at it.

  1. Open the menu (☰) and tap "Help" — the first row under Spaces. It is also the first row in Code's menu.
  2. Ask your question.

Places in its answers are links: each name in a path like Settings → Privacy & security → Lock screen is its own link, and tapping it shows that place and pulses it — "Settings" pulses the Settings button in the menu, "Lock screen" opens Settings at that row. A Space it names opens that Space; a message it finds opens the Space at that message.

To find where something was said, Help searches your messages on this device and shows you the places. The first time you use Help, your phone reads the older history of your Spaces in the background and keeps a copy for searching (it stays on the device, sealed like the rest when storage protection is on), so the first answers may cover only part of it. It never reads the messages' words back to the assistant. It also finds images and files by their names, a photo by what it shows, and the Code sessions this device has opened; a Code session it names opens in Code, on its computer.

Help is free for up to 50 questions a day; if your account pays for its own use, a Help question is paid like any other message, with no daily limit. It is a confidential Space only you are in, and it is not listed among your Spaces. It can see the names and kinds of your Spaces, who is in each one (owners, admins, people invited by email, people who joined, and who wrote), who wrote each message, the devices signed in to your account, and this device's settings; it cannot change anything for you — it tells you where to do it. It only answers questions about the app.

Not available when: you are not signed in. After your 50 free questions it says "You have used today's 50 help questions. They reset at midnight UTC." If it says "Help is not available right now — try again in a minute", wait a minute and ask again.

Your Spaces list

The menu (☰) lists your Spaces, newest activity first.

  1. Tap the magnifier to search them by name.
  2. Tap "People" for your private address book. Each person shows whether they are on Spaces yet ("not on Spaces": they need to open Spaces and sign in once — invites you send when you create a Space reach them then; an invite from inside a Space asks you to invite them again after they have). To add someone, tap + and give a name and an email, an @username, or both (a username must belong to someone on Spaces). Tap a person to open their card: their name, email, @username and phone, with "Message" (your chat with them, or a new one) and "Call" when you saved a phone number. "Edit" at its top right changes the contact, and deleting it is inside the edit card. In a verified Space, someone you saved here with a name shows under that name on their messages; tap the name to see their email or @username. Your names stay on your devices.
  3. Long-press a Space to pin, rename, archive, export or delete it. Archived Spaces are under "Archived" at the end of the list.

A Space listed as "Space" is one whose name had not reached this device; the app asks the Space for its name in the background and the row updates by itself.

Getting around: tap ✕ to close a card, or on a phone pull it down from the top. A card opened from another card (Settings → Back up & recovery) shows ‹ instead, and takes you back to the one you came from. "People", "New Space", "Invite to Spaces" and "Open a shared link" close the same way, back to your Spaces.

Short notes about something you just did ("Backed up", "Couldn’t save the contact") appear above the message box and fade on their own. One that needs you — your sign-in expired — stays, with its button, until you answer it.

To clean up: Settings → Preferences → "Remove empty Spaces". It lists the Spaces with no messages and nobody else in them, asks once, and deletes them on the server and from all your devices. A Space with even one message, or another person in it, is never removed.

Start a Space

A Space is a chat room with an assistant in it. Start one just for you and the assistant, or with other people — you can add people later too.

  1. Open the menu (☰) and tap "New Space" — or tap ✎ at the top of any room.
  2. The assistant: tap "Assistant" to choose which one joins, or "New assistant" in the list to give it your own name and instructions (how it should answer, what it should know). To change or delete one of your saved assistants, touch and hold it in the list (right-click on a computer) → "Edit" or "Delete" (on a slide).
  3. People (optional): tap "Add people" and pick from your contacts, or type an email or an @username. Once anyone is added, that row shows their names and how many; tap it to see everyone, hide or show each address, remove someone, or add more.
  4. A name: a Space with other people needs one — type it in "Name this Space". On your own there is no name field: the name follows the conversation until you rename it. With "Verified" on, exactly one other person and no name, the Space is a direct chat with them.
  5. Verified (optional): turn on "Verified" — only the people on the list can read and post; each proves their email with Google or an emailed code first. On your own, turn the same switch on for a verified Space of your own — you need one to use connections. You need to be signed in for this.
  6. Optional: pick a "Model" for the assistant.
  7. Optional: "Advanced" — - "Confidential tier": the assistant runs in a verified secure environment, so not even Witbitz can read the Space. - "Temporary": the Space is deleted when it is not used, and can’t be recovered. - "Private assistants": each person also gets an assistant only they can see. - "Continue an existing chat": bring in a ChatGPT, Claude or Gemini share link, or a WhatsApp chat export.
  8. Tap "Create". The Space opens and you can start writing.

The people you added get it at once: it appears in their Spaces if they already use Spaces. A short notice says who was invited, and names anyone who couldn’t be — someone who hasn’t used Spaces yet (in the hosted app, hasn’t opened it yet); invite them from ⋯ → "Invite someone", which has the link to send. To add people, rename, or change who can post later, use ⋯ in the Space.

To bring in an old conversation: "Advanced" → "Continue an existing chat", then paste a ChatGPT, Claude or Gemini share link, or pick a "WhatsApp chat export" file or a "Witbitz export (.json)" (a Space you exported earlier). A WhatsApp chat shows how many messages it holds and from when; choose the date to bring it in from, then ✓. Its documents and photos come in where they were sent when the export was made with "Attach Media" (the .zip); voice messages and videos are named, not brought in.

Not available when: you added people but left the name empty — the name field is marked and says "A Space with other people needs a name — everyone in it sees it." A verified Space needs you to be signed in.

Invite people

Anyone in a regular Space can share its link. The person opens it, gives a name, and is in.

  1. In the Space, tap ⋯ (More).
  2. Tap "Invite someone" to share a link by WhatsApp, Email, "Share…" or "Copy link" — or "QR code", which shows the link as a code to point the other phone's camera at (tap it again to hide it).
  3. Or open ⋯ → People → Invite by email and choose people; the Space appears in their Spaces when they open the app.

To open a link someone sent you: menu → "Open a shared link", paste it, tap "Open".

To invite someone to the app itself (not a Space): menu → "Invite to Spaces".

Not available when: the room is a direct chat (adding a third person would make it a group), or it is your private assistant room.

Verified Spaces

A verified Space lets only the people on its list read and post. Each person proves their email with Google sign-in or an emailed code. You must be signed in to make one.

  1. Start a New Space and tap "Add people".
  2. Turn on "Verified".
  3. Optional: turn on "Show @usernames" to show posters by their @username instead of display names.
  4. Name the Space and tap "Create".

In "Add people" you can type emails, @usernames, or "@company.com" for a whole domain. An @username is only used to find the person: the Space keeps a hash of their account, never the username, so others in the Space don't see it until that person opens the Space.

The person who creates a verified Space is its owner.

Reading needs the sign-in too, not only posting: a verified Space opens for you after a short "Sign in to this Space" card ("Google" or "Email me a code"). One sign-in covers all your verified Spaces on that device and renews by itself for 30 days; "Email me a code" gives a sign-in that renews quietly every time. A Space you made verified on one device is marked verified on your others too (the ✓ in the list), so they sign in for it before opening it.

Not available when: you are not signed in ("Sign in to create a verified Space — you need to be on its list yourself.").

Confidential Spaces

A confidential Space runs its assistant in an attested secure enclave, and the model runs on confidential hardware. The server can never read it. It uses an open model. The room shows an "ENCLAVE" badge at the top.

  1. Start a New Space and tap "Advanced".
  2. Turn on "Confidential tier".
  3. Go back and tap "Create".

To see how a confidential Space is protected, tap the "ENCLAVE" badge, or open ⋯ → This Space → Confidential tier.

New enclave builds Witbitz publishes are accepted for you automatically — Settings → Privacy & security → "Accept new Witbitz builds automatically" is on unless you turn it off, for your whole account. With it off, a new build makes your confidential Spaces wait and that sheet shows "Accept this build". Accepting is your own choice: it counts for all your confidential Spaces, on all your devices, and changes nothing for anyone else. The first build you meet is accepted for you.

If you see "This confidential Space is asleep", its secure environment was asleep; while it starts, the banner says "Launching a private enclave for this Space…". Tap the message box to wake it — it takes about 2–3 minutes, and the message bar says "Starting the enclave…" meanwhile.

Temporary Spaces

A temporary Space is deleted when it goes unused and cannot be recovered. Nothing is saved on your device. It shows a "TEMPORARY" badge.

  1. Start a New Space and tap "Advanced".
  2. Turn on "Temporary".
  3. Tap "Create".

Tap the "TEMPORARY" badge to read what it means.

Not available when: "Private assistants" is on — turning one on turns the other off. A temporary Space also cannot get a private room. When one has lapsed you see "That temporary Space has lapsed — it was wiped."

Direct chats

A direct chat is a verified Space with exactly one other person and no name. Each side sees it under the other person's name. You can have several direct chats with the same person.

  1. Start a New Space, tap "Add people" and pick one person.
  2. Turn on "Verified".
  3. Leave the name empty (the field shows the person's name), then tap "Create".

From a Space's member list: ⋯ → People → Who's been here (in a verified Space its sheet is titled "Members"), then tap a person and "Message" on their card — it opens your chat with them, or a new one if you have none.

If you type a name, it becomes an ordinary verified Space for two instead.

Not available when: renaming — nobody renames a direct chat. You also cannot invite more people into it.

Rename a Space

A Space alone with you is named automatically from the conversation until you rename it. A group is named by hand. A rename shows as a line in the chat and notifies the members.

  1. Tap the Space's title at the top, then "Rename".
  2. Or open ⋯ → This Space → Rename.
  3. Type the new name and tap ✓ (✕ cancels).

You can also long-press a Space in the menu list and tap "Rename".

Not available when: the Space has owners or admins and you are not one of them ("Only this Space’s admins can rename it."), or it is a direct chat.

Members, owners and admins

In a verified Space, owners and admins manage the list. An owner is also an admin. Admins can add people, remove members and make members admins; only owners can change owners or remove an admin.

  1. Open ⋯ → People → Who's been here (in a verified Space its sheet is titled "Members").
  2. The card lists owners and admins first (with "Owner" or "Admin" on the right), then you and the members, then people invited who have not opened the Space yet. Above them: "Add members" (owners and admins) picks people from your contacts, and "Invite via link or QR code" shares the Space's invite.
  3. Tap a person to open their card: their picture, @username and email, then "Message", "Add to contacts", and — for owners and admins — "Make admin" / "Remove as admin", "Make owner" / "Remove as owner" (owners only) and "Remove from this Space".
  4. "Remove from this Space" starts only on a slide (drag its icon to the right); the other changes ask you to confirm.
  5. "Rotate the key" (owners and admins), at the top of the card beside "Add members": the same members, a new key — for a lost phone or a link sent to the wrong place. Everyone in it gets the new key by itself and the conversation goes on; a copy of the old key opens nothing written after.

The change happens in place: the same Space, its history and its link, and a new key that only the people still on the list get — their devices take it by themselves. A removed person keeps what they had already read and can read nothing written after; their copy says "This Space continued without you, so you can’t post here any more".

In a regular (link) Space, the person who made it may see ⋯ → People → Manage members: remove people or add new ones, then "Apply changes". It changes the same Space in place — a new key that only the people who stay get, their devices take it by themselves — and gives you a link for each new member to send them.

Not available when: you are not an owner or admin ("Only this Space’s owners and admins can change its members.").

Talking to the assistant

Type in the message box and tap send. The assistant answers in the chat.

Choose how the assistant replies with the mode chip in the composer. The mode is shared by everyone in the Space.

To address the assistant:

  1. Type "@" and pick its name from the list, or tap "@AI" to send your message addressed to it.

Choose the model with the model chip. Some models show an "Effort" row — how hard the model thinks before it answers.

Change the assistant's name and instructions: ⋯ → Assistant → Assistant instructions, then "Save". Everyone sees that they changed.

Long-press a message for a row of emoji to react with (tap one again to take it off), and "Reply", "Play aloud" and "Copy message". In a confidential Space an assistant reply also has "Turn certificate".

Not available when: "@AI" is hidden in "Always" mode, since every message already gets a reply.

Private assistant panel

With "Private assistants" on, each person has their own assistant that only they can see. It reads the shared Space and helps you privately.

  1. In the shared Space, tap "🔒 Assistant" to open your private panel.
  2. Talk to it there; the shared room does not see this.
  3. To have it write something for the shared room, type your idea and tap "✍️ Draft".
  4. Drafts wait under "Awaiting your approval" in the shared room until you approve them.

Reminders: if something your message started waits for your approval and nobody answers, the device you sent it from gets a notification — "Waiting for your approval" once, 10 minutes after the turn (a Space), or "Code is waiting for your approval" after 1 and 10 minutes (Code). A tap opens the card in its Space, or Code on that computer and session. Nothing is sent once it is decided, and with "Show message preview" off the notification says only what it is. On a device with "Also protect this device’s storage" on and "Show message preview" off, Code's reminder cannot even tell which computer it is from: a tap opens Code rather than the waiting session, because nothing that could read it is kept on the device. Notifications must be on.

In a verified Space, your private assistant is verified as well — for your email only, never the Space's other members — and it uses the Space's sign-in, so it never asks you separately. In a confidential Space it is confidential too.

To add a private room to a Space that has none: ⋯ → Private room → Add a private room.

Not available when: the Space is temporary. The private room itself cannot be shared or invited to.

Meetings

  1. ⋯ → People → Meeting link: paste a Zoom, Meet or Teams invite (the whole text is fine — the link is picked out). Everyone in the Space then sees "Join meeting".

Siri

Say something to Siri, or share from another app (a place in Google Maps, a link), and it lands in a Space, posted as you — on an iPhone or iPad.

  1. Sign in with your email.
  2. Settings → Assistant → "Siri" → "Spaces" : switch on each Space Siri may write to. It goes by its own name in the list, and a Space with your private assistant brings it along as " assistant". To change a name, tap it on the line under that Space's name (Escape puts it back).
  3. On the same screen tap "Configure the Shortcut" → "Add Shortcut". Open it in the Shortcuts app and replace the line "PASTE YOUR SIRI KEY HERE" with your key (the copy icon next to Key) and "PASTE YOUR SPACE LIST HERE" with your Space list (the copy icon next to Space list). Need a hand? "Ask Help to set up Siri" on the same screen.
  4. Use it: say "Hey Siri, Send to Spaces", say the message, then pick the Space — or, in another app, tap Share → "Send to Spaces" → pick the Space.

It arrives in the Space marked 🎙 the next time you open Spaces — a verified Space, the next time you open that Space. Each person sets up their own Siri, with their own names for the Spaces. After switching a Space on, copy the Space list into the Shortcut again.

Share from other apps

Send a place from Google Maps (or a link, a photo's text, anything an app shares) into a Space.

Outside agents

Other AI agents can reach your assistant in a Space's private room.

  1. Open ⋯ → Assistant → Outside agents.
  2. Choose what outside agents see your assistant called and turn on "Let outside agents reach my assistant".
  3. What they send arrives in your private room; your assistant drafts the answer and you approve it before it goes back.

What an outside agent sends waits in your account's inbox, sealed for your devices, until one of your devices with the app open writes it into the room — usually within a minute. Nothing is lost while every device is closed; it waits up to 7 days.

Grok Bot is set up as a connection (see Connections → Grok Bot), not here.

Linking private rooms

Attachments, photos, voice and files

  1. Tap the attach button to choose photos, take a picture, or pick a file.
  2. Attached items show above the message box; tap one to preview, or remove it.
  3. Send the message. The assistant and everyone in the Space see them.

One message carries up to 4 photos and 2 files. Each file can be up to 3.5 MB. A long paste becomes an attached text file; you can turn it back with "Paste as text".

Press-to-talk:

  1. Tap the microphone.
  2. Speak; the words appear in the box as you talk.
  3. Tap stop to keep the words and edit them, tap send to send what you said, or tap ✕ to cancel.

The first use may download a voice model. Allow microphone access when asked.

Files and pictures the assistant makes open in a viewer, where you can save, copy or "Pin to views". Pinned items are in "Shared views" at the top of the room; each one's menu has "Rename", "Pin to top" (it then leads the list — "Unpin from top" undoes it) and "Archive". A red dot beside one means it changed since you last opened it; opening it clears the dot on all your devices.

Safe media

A Space can show photos and PDFs from other people only from safe copies. When someone posts a photo or a PDF, their device has a copy made inside an attested secure enclave: a parser that keeps nothing reads the file there and returns only its pixels, and the enclave turns them into a plain picture (for a PDF, its first page). The hardware signs that this enclave made exactly that picture. Before your device shows it, it checks the signature, so a file built to attack a phone is never opened by yours. A safe copy is marked "Safe copy".

  1. Open ⋯ → This Space → Safe media and turn it on. In a Space with admins, only an admin can change it; otherwise anyone in it can. Everyone sees a line saying who turned it on or off.
  2. The Space shows a "SAFE MEDIA" badge at the top.

A photo or PDF without a safe copy yet shows as a card with its name, size and who sent it. "Make safe" makes the copy for everyone in the Space. "Open original" opens the file itself on your device. If a safe copy exists, your device first checks that the original is the file the copy was made from. Your own photos and files show as usual. In a Space with Safe media on, other people's profile pictures show as their initial. With "Strict file viewing" on as well, "Show" opens the safe copy, not the original.

Notifications and message preview

  1. Open Settings → Preferences → Notifications and tap to turn on.
  2. Once on, choose "Show message preview": On shows the sender and first line; Off says only that something arrived.
  3. For one Space, use the ⋯ → This Space → Notifications switch: on, this Space notifies you; off, it stays quiet.

Each device keeps its own choices: the switch in Settings and each Space's switch apply to the device you set them on, and a Space you turned off stays off there. Your devices also look after each other: a new Space, or one you join, notifies every device of yours that has notifications on — including one that is closed — unless that device turned the Space off. Signing a device out in Settings → Privacy & security → "Your devices" stops its notifications at once.

The app's icon shows how many Spaces have something you have not read — the same Spaces the green dot on ☰ counts — and opening a Space clears its notifications and its share of the number. It needs the app installed with notifications on (on an iPhone, iOS 16.4 or later); a device with the lock's "Also protect this device’s storage" on updates the number when you open the app, not when a message arrives.

On an iPhone, install the app first (Settings → Preferences → "Install app"). With Lockdown Mode on, an iPhone gives web apps no notifications: the row says "Off while iPhone's Lockdown Mode is on" — installing again does not help; it works once Lockdown Mode is off.

Appearance. Settings → Preferences → "Appearance" opens a card for this device's visual style. Choose System, Light or Dark mode; System follows your device. Theme opens a list of Default, Sage (the trip demo's colors), Sand, Rose, Lavender, Ocean and Slate, each with an Aa swatch preview and light/dark palettes. Accent, Background and Text accept a color picker or a six-digit hex color. Custom colors are kept separately for each theme and light/dark palette. Font offers System, Rounded, Serif and Monospace. Changes apply immediately and survive reopening the app. Reset restores Default, System mode and font, and removes custom colors. These preferences also apply to the Code section, including an already open Code view, and in the hosted app its own screens too, such as the lock and sign-in, and change appearance only.

Default room. Settings → Preferences → "Default room" opens your defaults for new rooms, with the current choices already selected. Choose Enclave protection, Verified people, Private assistants, Temporary room (where available), Assistant and Model. "Show creation options" keeps the full create form; turn it off for just Add people and Create. Temporary rooms cannot have durable private assistants, so enabling either option turns the other off. Choices sync to your other devices through your encrypted account sync and apply when starting the next room; existing rooms are unchanged. Without account sync, they stay on this device until sync is available. Appearance is separate, saved per device and never changed by room defaults. Existing Simple setup users keep their prior room defaults.

Software update. Settings → Preferences → "Software update" shows this version and what the app is doing about the next one: up to date (with when it last checked), checking, downloading (with a percentage), "Download paused" (it goes on when you come back to the app), ready, or "Could not check" (you may be offline — "Check for updates" tries again). A blue dot on ☰ and on Settings means a new version is ready. "Check for updates" asks now; when a new version is ready, "Update now" switches to it at once. "Update automatically" is on by default: a new version downloads in the background and Spaces switches to it while you are not using the app. Turn it off and new versions still download (you see the progress), but the app moves to one only when you tap "Update now". Closing and reopening the app does not move it either. Version. Under "Version": Beta versions gets the newest build as soon as it is published, before the build enclave checks it — for trying changes early. It is on by default for now. Pin a version keeps this device on one released version: turning it on turns Beta off and shows the versions you can pick. With both off, the device follows the official versions, each rebuilt and checked by the build enclave. A device never goes back to a version older than one it has already run: turning Beta off waits for the next official version that is at least as new, and older versions are not offered to pin. Only official versions can join a Space that requires a checked app.

Not available when: notifications are blocked in your device settings ("Notifications · Blocked in device settings"), or the app is not installed on a device that needs it.

The lock and storage protection

The lock asks for Face ID / Touch ID to open Spaces on this device — on a computer, its own check: "Windows Hello lock" on Windows, "Touch ID lock" on a Mac.

  1. Open Settings → Privacy & security → "Lock screen".
  2. Turn on the "Face ID lock" switch.
  3. Choose "Also lock after being away" (only when the app opens, or after 1, 5 or 30 minutes in the background).
  4. Optional, once the lock is on: turn on "Also protect this device’s storage" (it says "Turn on Face ID lock first." until then) — everything Spaces keeps on this device stays encrypted until you unlock: your keys and messages, your name, contacts and to-dos, and Code’s computers and their keys. With it on, tapping a call notification opens the app rather than the meeting itself.

To turn it off, open the same card and turn the "Face ID lock" switch off. Its ‹ goes back to Settings.

When locked, tap "Unlock".

Not available when: the device has no built-in authenticator (the row is hidden).

Securing a computer: which passkey works

"Secure this computer" needs a passkey that can hand Spaces the same secret key every time it is asked — that key encrypts everything Spaces keeps on the computer. Many passkeys can sign you in but cannot keep a key. The setup TESTS the passkey (it asks it a second time) before anything is sealed; if the key does not come back, it says so and offers your phone instead.

What has been measured on a Windows 11 computer (25H2, October 2026):

To use Windows Hello in Edge: when Windows asks where to save the new passkey, choose this Windows device (Windows Hello). If it does not ask, open Windows Settings → Accounts → Passkeys → Advanced options and turn off Microsoft Password Manager (and 1Password) as passkey managers, then set Spaces up again. Windows Settings → Accounts → Passkeys lists the passkeys saved on the computer; their names say what each is for and which browser made it ("Spaces · this computer & backup · Chrome on Windows · date"). Do not delete them by guesswork: the one a browser's Spaces is sealed under cannot be replaced, and deleting it means starting that browser over.

The same applies to a backup passkey made on such a computer: if it cannot give the key back, it cannot restore your backup. The recovery code (with the email check) always can.

Devices, sign-in and recovery

Sign in with your email to keep your Spaces, sync devices and join verified Spaces. Back up so you can restore everything on a new device.

Sign in: Settings → Profile → "Login".

Turn on backup:

  1. Open Settings → Privacy & security → "Back up & recovery".
  2. Tap "Turn on backup". Where this device can use a passkey, it asks for one first — finish it to restore with your fingerprint or face later, or close it to use the recovery code alone. Then it shows your recovery code.
  3. Tap "Copy code" and keep it in a password manager; "Test this code" checks you saved it right.
  4. No passkey yet (or a new one)? "Add a passkey" (or "Replace passkey") does it later. Some passkeys ask twice: the second prompt ("One more step: confirm the passkey you just made…") reads the key back. A passkey is named for its purpose, device and day in your keychain, e.g. "Spaces backup · Chrome on Windows · 11 Oct 2026"; "Back up & recovery" says where your current one was made — only that one opens your backup, so older "Spaces backup" passkeys can be deleted from your keychain. - If a passkey can't hold your backup's key, it isn't added and the app offers "Use my phone": a code appears, scan it with your phone and confirm there. On Windows, a passkey saved to this Windows device works (Windows Settings → Accounts → Passkeys → Advanced options → "Save passkeys to this Windows device"); some passkey managers reached through the Windows box (1Password, for one) can't hold the key.
  5. Optional: tap "Protect my backup" so your recovery code alone can't restore it — a restore then also asks for your email or passkey. Once one is set, the row offers "Add a second recovery factor", so losing one can't lock you out.

If this device stopped backing up (for example after you turned backup off here), "Back up & recovery" offers "Turn on backup for this device".

Add a device:

  1. On the new device: Settings → Privacy & security → "Bring my Spaces from another device" — or, on the welcome screen of a new install, "Restore your Spaces" → "Bring them from another device"; it shows a code.
  2. On a device that has your Spaces: Settings → Privacy & security → "Add a device", and scan it. It asks for your passkey ("Confirm with your passkey…") before the keys leave, then says how it goes: "Sent — waiting for your other device to pick it up…", "Your other device has it — finishing there…" and "✓ Added — your other device is ready." If the new device does not pick it up, it says so: show a new code there and scan again.
  3. On the new device: it asks "Sign this device into ?" Continue only if that is your account — if you don't recognize the address, cancel: someone may have scanned your code. The new device gets your Spaces and your sign-in, so it posts in verified Spaces without signing in again. If the keys come over but your Spaces don't, both devices say why (a protected backup, no backup saved yet, or keys that don't open it) and what to do.

Restore: on a device with no Spaces, "Restore your Spaces" on the welcome screen, then "Use a passkey", enter your recovery code and tap "Restore my Spaces", or "Bring them from another device". There is no Restore in Settings: on a device that has your Spaces they stay in sync by themselves. To bring a different account to this device, reset it first (Settings → Reset this device).

See and sign out devices: Settings → Privacy & security → "Your devices". "Sign out" on a device, or "Sign out all other devices" (with more than one other): each clears itself the next time it is online, and waits under "Waiting to sign out" until then. This device stays signed in.

See your recovery code again: "Back up & recovery" → "Show recovery code". To stop backing up on this device: "Pause backup on this device" — your Spaces stay here and keep working, but nothing from this device is backed up or synced to your other devices until you tap "Turn on backup for this device". The encrypted backup stays on the server, so your passkey or recovery code can still restore it.

Keep an offline copy: "Back up & recovery" → "Download an offline copy".

If a device that holds one account's Spaces is signed in as another, nothing from it is added to the second account, and it asks: "This device's Spaces are not being backed up to …" — "Back them up to …" if they are that account's, or "They’re not mine", which offers "Switch account" (it clears this device; the first account's Spaces stay in that account and on its other devices).

Log out: Settings → Profile → "Log out" (it shows your email beside it). On a device that has Spaces but is not signed in, the same row reads "Switch account". Logging out clears this device; with backup on, your Spaces come back when you log in again. Delete everything: Settings → Profile → "Delete my account".

Choose a public @username: Settings → Profile → "Choose a username". Its card also has "Show my name to people who look up my username", "People can find me by my email address" (off: only by your username — an invite to your email then arrives as an emailed link) and, once you have one, "Release my username".

Connections

A connection lets an assistant use another service for you — your calendar, an issue tracker, a weather API. Its key or sign-in is kept in a sealed enclave and never reaches the Space, the server or the model.

How it is set up — two levels:

Your connections — Settings → Assistant → "Connections". "Connected" lists every connection you have, each with how many Spaces use it; tap one for its card, where "Disconnect" deletes it for every Space at once. With many, "Search your connections" narrows the list.

Which Spaces use a connection — on its card, "Spaces". Every verified Space of yours, each with a switch ("Find a Space" narrows a long list): turn it on to let that Space's assistant use the connection, off to stop. It works from Settings or from any Space's Connections. A Space with admins lets only its admins change its connections; it says so on its line.

A Space's connections — in the Space, ⋯ → Assistant → Connections. "On in this Space" lists what this Space's assistant may use; "Your connections" lists your others, each with "Add" to turn it on here (a service comes with its recommended tools). Tap a connection for its card: "Stop using here" turns it off in this Space only, and "Includes" chooses which of a service's tools the assistant may use. The recommended setting offers every tool; the ones that only read run without asking, and anything that may change something asks you first, as a card. Change any tool and the tools card shows "Custom setting", with "Use recommended" to go back.

To connect something (the same list in both places; from a Space it is also turned on there at once):

  1. Sign in with your email.
  2. Find the service in the list (by category, or search) and tap "Connect": its own sign-in opens, and when you come back it is connected.
  3. A service that is not listed: under "Other ways to connect", "Other MCP server" takes its address, an optional name, and a token or nothing — with no token, "Connect" opens the server's own sign-in.
  4. An API key instead: "API key" (also under "Other ways to connect") takes a name, the host it is for and the key; tap "Attach".

Grok Bot. Grok Bot works both ways: your assistant sends a Grok bot jobs (it does them on its own computer), and it sends results back into the Space. Each Space gets a Grok bot of its own, and you set it up by talking to Grok Bot.

  1. In the Space: ⋯ → Assistant → Connections → "Grok Bot" (under Agents) → "Create the message". It makes this Space's connector address and one message for your main Grok bot, with a Copy button.
  2. Paste the message to your main Grok bot. It adds a connector named after this Space ("Witbitz · "), creates a new bot with the same name, and sends that bot the routine's instructions: do each task, ask you before booking, paying, deleting or messaging anyone else, note each run in memory, and always send the result back with that connector's message_member tool. Once the routine is saved, the new bot shows you links to its Webhook URL and its key — you need both in step 3. If your bot cannot add the connector itself, the card shows the name and URL to add by hand: in Grok Bot, Connectors → New Connector → Custom. When the new bot saves its routine, Grok Bot stops at a "Review an action" card ("Save the routine …", flagged by its Auto-review): open that bot's chat and tap "Always allow" (the routine is saved only once you allow it).
  3. In Grok Bot, open the new bot → its routine, and copy its Endpoint (or Webhook URL, api2.cursor.sh/automations/webhook/…) and its key — only the app shows them, a bot cannot read them out. In Spaces, paste the Endpoint into the card's "Endpoint (or Webhook URL)" field and the key into its "Key" field, then tap "Connect". The key stays sealed in the Sentinel; paste it only here.
  4. Ask your assistant in that Space to send its Grok bot a job. You approve the request; the bot runs it and its result arrives in the Space's private assistant (or the Space).

Grok Bot then shows as On in that Space's Connections ("Answers this Space's private assistant"). Each Space's bot is separate: setting up another Space never touches this one. If you lose the connector address, tap Grok Bot there → "Set up again" — it asks first, because the old address stops working and the main bot needs the new message. "Stop using here" ends this Space's bot and its key.

Every job your assistant sends already ends with how to send the result back, and asks the bot to note the run in its memory. If a result never comes back, the routine finished without sending it: add to the routine's instructions "Your last step, always: send the full result back with message_member." A chat with your bot may not see what a routine did — look in the routine's own run history. An older routine made on grok.com (an "Endpoint" on grok.com/webhook/automation/… with a "Signing secret" whsec_…) still works the same way: paste those two instead.

Claude Code. Claude Code works the same way as Grok Bot: your assistant sends it jobs, and it sends results back into the Space. A job runs as a Claude Code scheduled task in Anthropic's cloud, so no computer of yours needs to be on. It needs a claude.ai plan (Pro, Max, Team or Enterprise), and runs count against that plan. Set it up from a computer: the task's API trigger can be added only on claude.ai in a computer's browser, not in the Claude app on a phone. Each Space gets a task of its own:

  1. In the Space: ⋯ → Assistant → Connections → "Claude Code" (under Agents) → "Make this Space's connector". At claude.ai/customize/connectors → Add custom connector, add it with the name and URL the card shows.
  2. At claude.ai/code/routines → New task: paste the prompt the card shows, choose the repositories it may work in, add an API trigger, and keep only the connectors it needs (this Space's among them). The task runs without asking you, with whatever you give it.
  3. On the task's API trigger, copy the Fire URL, then Generate token (it is shown once). Paste both on the card and tap Connect.
  4. Ask your assistant in that Space to send Claude Code a job. You approve it; the card says it leaves this Space's encryption, because Anthropic sees the job and the code it works on. The result, with a link to any pull request, arrives in the Space's private assistant (or the Space).

Each task can be fired 30 times an hour. If a result never comes back, open the run at claude.ai/code/routines: the task may have stopped to ask you something, or ended without sending it.

Muse. Meta's Muse works with one of your assistants: it messages it, reads its answers, and asks it for jobs. Nothing can start Muse from outside, so a recurring task in Muse asks your assistant for work.

  1. Open a verified Space, then ⋯ → Assistant → Connections → "Muse" (under Agents) — or Settings → Assistant → Connections → "Muse", which asks which of your verified Spaces it works with. Muse works with that Space's private assistant (or the Space itself), one at a time: connecting it to another Space moves it there.
  2. Spaces shows a card with a message to send Muse, a key, and an address with the key in it. Send Muse the message; when it asks for the key, paste it into Muse's secure credential prompt, not the chat.
  3. Muse adds the connector and sets up its recurring task. Each answer it sends comes to your assistant, and you approve what goes back. To stop it, remove Muse under ⋯ → Assistant → Outside agents.

Dots (OpenAI). Your dot (in ChatGPT) works with one of your assistants the same way: it messages it, reads its answers, and asks it for jobs.

  1. Open a verified Space, then ⋯ → Assistant → Connections → "Dots" (under Agents) — or Settings → Assistant → Connections → "Dots", which asks which of your verified Spaces it works with.
  2. In ChatGPT add an app: Add → Create custom MCP server, name it, paste the Server URL from the card (it holds the key — keep it private) and choose no authentication. Then send your dot the message from the card.
  3. On some ChatGPT plans custom apps can only read: if your dot cannot send messages to your assistant, it needs a Business or Enterprise plan. To stop it, remove Dots under ⋯ → Assistant → Outside agents.

A connection made on another device appears on this one once your account has synced ("attached on another device — it syncs to this one").

Not available when: the Space is temporary (its settings are deleted with it), you are not signed in ("Sign in with your email to add a connection."), or the Space is not verified — in an open Space anyone with the link could have the assistant spend your connection, so the sheet only says "Connections work only in a verified Space…" and offers nothing to connect there. A Space of your own can be verified: in New Space turn on "Verified" with nobody added.

Mail

Forward email into a Space, for the assistant to read when you ask — in the Space, ⋯ → Assistant → Mail. It shows only in a verified Space that has you as its only member (mail is yours alone, so it never lands where others read), when you are signed in and the Space is not temporary.

  1. Open Mail. The Space gets its own address (…@mail.witbitz.chat) — tap "Copy address".
  2. In your mail app, make a rule that forwards to that address only what the assistant should see (one sender, a label, a subject). Forward a mail by hand the same way.
  3. Ask the assistant about it ("what did the airline send?"). Mail never appears in the conversation; the Mail card lists what came in — tap one to read it. Documents and images attached to a mail come in too, and the assistant can read them.

Who can read it: mail to the address is received inside an attested enclave and sealed to this Space there — Witbitz cannot read it. It must arrive encrypted (Gmail's does); a sender that cannot is refused. A mail is written by its sender, not by you, so the assistant treats it as information, never as instructions.

"Stop receiving mail" ends the address: mail sent to it bounces from then on. "Get a new address" starts again with a different one.

Code and computers

Code lets a coding agent work on your own computer, driven from your phone. It is free to try: every account gets 100 credits of models and 25 credits of tools a month on one computer; invited accounts get 5,000 and 500 a month on up to three — in each app separately (an account in Witbitz Spaces and one in the hosted app each have their own). When a month's allowance is used up, Code says so and when it comes back (the 1st of the month), with a link to your plan.

To install it, start at witbitz.chat/code?app=spaces in Witbitz Spaces, or witbitz.chat/code in the hosted app — open that page on the computer you want to use: it shows the one line to paste for that computer, so it can be copied rather than typed.

  1. Open the menu and tap "Code".
  2. With no computer paired, Code says "Connect a computer to use Code" and opens the setup steps: pick Windows, Mac or Linux and follow them on that computer (or tap "Send to my computer" to send yourself this app's link; the same steps are on that page).
  3. Follow the steps the computer shows to pair it with your account. Once paired it appears in Code's list of computers.
  4. The same steps are always on your phone: in Code, open its menu (☰) → Computers → "Add a computer" — use it for another computer, or to see the steps again.
  5. To start working: in Code, open its menu (☰) and tap "New session", then pick the folder on that computer it should work in (or make a new one). Your sessions are listed in the same menu.

The steps on each computer (as "Add a computer" shows them):

The composer's mode chip in Code chooses how much the coding agent may do without asking. A new session starts in Auto: a reviewer on your computer decides each request, and only what it is unsure of comes to you as a card. Switch the chip to Manual (every request asks you), Accept edits (file edits go through, secrets still ask) or Plan.

In Code, the menu (☰) picks the computer when you have more than one, and lists that computer's sessions with a search and "New session" (pick the folder it works in). The model chip chooses the model and, where the model offers it, its Effort (how much it thinks first). While the agent is working you can still type: your message is sent into the running turn and the agent reads it at its next step; with the box empty the button stops the turn instead.

While the agent works through a plan, its task list is pinned above the conversation; when it moves on, the list stays with the reply that made it. To remove a computer: Code's menu (☰) → Computers → tap the computer → "Remove" — it starts only on a slide and then asks, because the computer’s coding keys are revoked and it stops working until you pair it again. To delete a session: touch and hold it in Code's menu (or right-click it), then "Delete session" — also on a slide; it is removed from the computer for good.

Your computer's security (Windows, witbitz-code 1.7 or later): the computer checks its own security settings — Microsoft Defender and its protections (real-time, tamper, cloud, blocking unwanted apps and malicious sites), memory integrity, the firewall, disk encryption (BitLocker), Windows updates, the admin prompt (UAC), and whether you use an administrator account every day. On the computer, Start menu → Witbitz Code shows the list with what to change; "Fix" changes a setting after Windows asks for permission, and "Open in Windows" opens the Windows page for a setting Windows checks itself (memory integrity needs a restart). From your phone: Code's menu (☰) → Computers → tap the computer → "Security" shows the same list (the fixes are made on the computer). It is a check of settings, not a malware scan — it finds weak settings before something gets in, but malware already running on the computer could hide from it. Microsoft Defender is enough as the antivirus; the most useful extra steps are usually turning on blocking unwanted apps, network protection and memory integrity, and using a standard (non-administrator) account day to day.

Files in a session

To put a file on the computer or take one off it: in a Code session, tap the folder button at the top right (Files). It lists the session's folder — tap a folder to open it, "‹ Up" to go back. Tap a file to download it: on a phone the share sheet opens ("Save to Files", AirDrop…), on a computer it downloads. "Upload here" sends files from this device into the folder you are looking at; when a file of that name is already there it asks: "Replace", "Keep both" (saved as "name (2)") or "Skip". Up to 20 MB a file. Only the session's own folder is reachable, and a file that looks like a secret (a .env, a private key, anything in .git or .ssh) is never downloaded this way. The button appears where the computer's witbitz-code supports it — on an older one, update it (the note at the top of Code's list says how). To hand a file to the agent with a message instead, use the composer's "+" (Attach files).

The agent's browser

The coding agent can have its own web browser. While it is open, a globe at the top of Code opens it on your screen — from your phone and your computer at the same time; each sees it laid out for its own screen. The dot at the top left says how the view is: green live, amber connecting, red something went wrong.

The buttons, left to right (each appears only when it applies): the hand takes the controls, so you can tap, scroll and type in the agent's browser yourself; ✕ closes the view and goes back to the app (the browser keeps running); the keyboard button opens your keyboard while you hold them; the speaker turns the browser's sound on or off on this device only (it appears when the page plays sound; not on an iPhone in Lockdown Mode, which turns web audio off); the arrow gives the controls back to the agent; the box with an arrow offers the page the agent's browser shows: "Open in browser" opens it in your own browser — on your own connection and with your own sign-ins, so a site that refuses the agent (YouTube, a Google sign-in) works there — and "Copy link" copies its address, to paste into an incognito window; while you hold the controls, the glasses open an incognito window on the page you were on — a fresh session with none of the browser's sign-ins, which the agent cannot see and which is thrown away when you close it or give the controls back; the screen-or-phone button switches between the desktop and the mobile version of sites, for everyone watching (the page reloads in that version; desktop is the default, whoever holds the controls); the corners make the view fill the screen (✕ in the corner leaves); the power button closes the browser itself — tap it, then again while it is red — its tabs and sign-ins are thrown away, and the agent opens a fresh one when it needs it (the agent can close it too, but not while you hold the controls); the question mark opens Help.

The globe opens the browser in a panel inside the app, on a computer and on a phone (✕ closes it; Escape too on a computer).

On a phone, Spaces stays upright — turned sideways it asks you to turn it back — except while you watch the agent's browser: then turn the phone sideways for a wider view.

To see the page in the chat, ask the agent for a screenshot ("show me the page"): the picture shows under its answer — tap it to see it large or save it. It is kept in Witbitz's own folder for that session, never in your project. On your own computer this needs witbitz-code 1.13 or later (1.15 from the Windows installer or pipx).

While you hold the controls the agent cannot see or use the browser. Anything you log into there, though, it can use afterwards — so do not type a password you use anywhere else.

On your own computer. With witbitz-code 1.8 or later, the agent on a paired computer has a browser too, watched and taken over from the app the same way. It is on by default and needs Node.js on that computer; to turn it off, start Code with witbitz-code serve --no-browser (or set WB_BROWSE_DISABLE=1). There the agent runs as you, so holding the controls asks it to wait rather than locking it out. In Code, ☰ → Computers, that computer's card shows "Private browsing: off"; tap it to turn it on, and its browser forgets cookies, logins and history every time it opens or closes, as on a cloud computer; witbitz-code serve --private-browser turns it on from the computer. It needs witbitz-code 1.11 or later (1.13 from the Windows installer or pipx) — an older one shows no switch — and the Linux browser card's full agent browser does not offer it yet. On Linux (witbitz-code 1.9 or later), the Witbitz Code page on that computer has a Linux browser card: "Install it" downloads the full agent browser — the same Chromium, sound and take-over as Witbitz's own (about 650 MB, checked before it is unpacked) — and the agent uses it from then on. Nothing downloads until you press it; a newer version shows there as an update.

When the agent's browser runs in a data centre, some sites limit what they show to it: video sites such as YouTube treat it as a bot — "Sign in to confirm you're not a bot", and videos that do not play ("You're offline" is the site's message, not the app's). Open such a page in your own browser with the arrow button.

Cloud computers

A cloud computer is your own computer in the cloud for the coding agent — nothing to install, and it works from your phone alone. It is a protected virtual machine only you can open: Witbitz cannot read its files. Its files are saved encrypted after every turn; it sleeps after 30 minutes unused and wakes when you open it (the first open after a sleep takes a little longer). The agent's keys stay out of it: a separate protected machine, its key guard, adds them to each call.

Cloud computers are part of the Plus plan. During early access they are open to accounts Witbitz has invited; on any other account, "Create" says "cloud computers are not open to this account yet".

The hours a cloud computer is open are counted in credits — about 23 credits an hour — against a monthly allowance: 5,000 credits (100 on the free allowance). Settings → Profile → "Your usage" shows this month's total and when it starts again (the 1st). When the allowance is used up, the computer saves and closes, and opening one says so until the month starts again.

  1. In Code, open its menu (☰) → Computers → "New cloud computer".
  2. Give it a name. A GitHub token is optional — add one to clone, push and open pull requests ("How to make the GitHub token" shows the steps).
  3. "What its own network may reach" chooses the internet the agent can use directly: "Nothing" (the default — no outside access), "Packages and GitHub" (package registries and GitHub, so it can install dependencies and use your repositories), or "Hosts I choose" (one hostname per line). Its browser is different: it reaches the whole web in every mode, and the agent can use the browser — so through the browser the agent can reach the web too.
  4. "Private browsing" makes its browser forget cookies, logins and history every time the computer opens or closes.
  5. Tap "Create". Its model keys come from your Spaces account, like a paired computer's: up to 5,000 credits of models and 500 of tools a month, and it counts as one of your three coding computers. Its hours use your cloud computer allowance (below).
  6. Save the key file it offers ("Save the key file") and keep it like a key: it brings the cloud computer back if you remove it or move to another account ("Restore from a key file", in the same Computers list). Whoever has the file and can sign in to Witbitz can take the workspace over.

It then appears in Code's list of computers like a paired one: pick it and start a session. Its agent has a browser too (above); while you hold the controls of a cloud computer's browser the agent is locked out of it, and the whole web is open to you there — but the agent can use whatever you log into afterwards.

To stop it: Code's menu (☰) → Computers → tap it → "Close" — a last save, then it shuts down; your files stay for the next open. "Remove" (on a slide, then asked) closes it and forgets it on your account; only its key file brings it back.

Usage

  1. Your own usage across Spaces: Settings → Profile → "Your usage".
  2. Usage in one Space: ⋯ → Assistant → Usage; "Export CSV" saves it.
  3. The owner of a verified Space can pay for members' use with ⋯ → Assistant → Fund this room — while they are its only owner and admin. Only they can then change its members; the funding stays with the Space when they do.

Plans and pricing

The Spaces website, getspaces.witbitz.chat, has the plans: Home (what Spaces is), Pricing, and Account (your plan and this month's use, after signing in with your email).

Early access: every account is free right now — nothing is charged. The Account page shows what your use would cost.

  1. Free — $0 a month: 50 credits of AI assistant use each month; Code on one computer with 100 credits of models a month; unlimited Spaces, people and messages.
  2. Plus — $10 a month: 800 credits of AI assistant use each month, a cloud computer for the coding agent (see "Cloud computers" — during early access, on invited accounts), and everything in Free. Free during early access.
  3. Credits: packs of 500, 2,000 and 5,000 credits ($5, $20 and $50) for use beyond a month's allowance. The allowance is used first, then credits; credits don't expire. Available when early access ends.

Use is counted in credits everywhere in the app — Your usage, a Space's Usage, funding a room, Code: 100 credits = $1.

Only the AI assistant is metered — Spaces, messages and everything between people never are. Use is what the AI provider recorded for you (every account has its own keys), and it is the use of the person who asked, in any Space. If you run out, the assistant pauses for you with a clear message; your Spaces keep working. To see your use in the app: Settings → Profile → "Your usage".

Settings overview

Open the menu and tap "Settings". Rows are in four groups.

In a Space, ⋯ has four groups too: People, Assistant, Private room and This Space (Rename, notifications, "Export (JSON)" to save its messages, "Delete Space"). "Install app" appears there when the app can be installed.

In the menu list, long-press a Space for "Pin", "Rename", "Archive", "Export (JSON)" and "Delete".

Troubleshooting

Machine-readable source: guide.md · Generated 2026-10-27T07:05:42Z · build f301388c · every doc in one fetch: llms-full.txt (HTML)