Live feed

The live feed puts what your customers write — messages on your channels, comments, reviews, marketplace questions — on a screen as it arrives. Use it on a TV at an event stand, as a review wall in a lobby, in a shop window, or for questions from the audience at a talk.

Everything that reaches the screen has passed your filters first: AI moderation keeps unsuitable content off, and the privacy filter hides who wrote it. The screen opens through a screen link and needs no signed-in panel, so the TV never holds anyone's session.

You find it in the sidebar, right under Dashboard. If your plan or your role does not include the live feed, the entry does not appear.

Boards

A board is one screen's worth of settings. You can run several at once — a different event in each city, a lobby wall and a stand TV — and each has its own channels, keywords, dates, filters and screen link.

Choose New board, give it a name, and save. Every filter starts at its safe setting: moderation on, privacy on, at most three items per person per ten minutes.

What a board collects:

• Messages and/or comments, reviews and questions.
• Channels — pick WhatsApp, Instagram, web chat and so on; none selected means all of them.
• Keywords or hashtags — only items containing one of them are shown, for example #expo2026. Empty means everything.
• Starts / Ends — outside these dates the board collects nothing.
• Board is on — the kill switch. Off stops collecting and closes the screen link at once.

Your own team's replies never appear. With Show Wiz's replies on, the AI's answer is shown under the message it answers, with the same filters.

Changing what a board collects also changes what it already shows. Narrow the channels, kinds or keywords and the items that no longer fit leave the board at once — the screens, the moderation queue and the counters — without reloading the TV. Widen them again and those items come back as they were (on screen, held, waiting), as long as they are among the board's 100 newest items. Nothing is deleted.

The time on each item is when it reached you. A sender whose clock is wrong — an e-mail dated hours ahead — never shows as "in 4 hr.".

Filters

AI moderation

Each item is checked before it is shown: swearing, insults, sexual content, personal data in the text, spam, and — in images — nudity, violence, a child, or someone's ID card, bank card or other private document.

Moderation is fail-closed. If the check cannot run — the AI is not set up, the wallet is empty, the service does not answer — the item is held, not shown, and waits in the moderation queue for a person. An image that could not be checked is left off the screen even when the text is fine.

A held item says why the check could not run — for example the AI provider rejected the request, the wallet balance is too low, the daily moderation limit is reached. The check is tried again automatically a few times over the next hours (after 5, 10, 20, 40, 80 and 160 minutes); once it works the item is shown or held for a real reason. Retry moderation in the queue tries all of them again at once.

The check uses your workspace's AI, one short request per item, and is counted under live feed moderation in Usage. When the same message lands on two boards, it is checked once.

Privacy filter

With privacy on, nothing on the screen says who wrote it: no name, phone number, e-mail, @handle or profile photo. Senders appear as Visitor #4821 — the same number for the same person on that board — or simply A visitor. Phone numbers, e-mail addresses, @handles and the sender's own name inside the text are replaced with •••.

Switching privacy on later also applies to what is already on the board.

Flood limit

Items per sender in Per minutes — for example 3 in 10. Further items from the same person are not shown; with Count the extra ones on, their last item says +2 more instead.

Approve before showing

Every item waits in the queue until a moderator approves it, even when moderation passed it.

Moderating during the event

On the board's page, the left side shows everything that came in with its status — On screen, Awaiting approval, Held, Checking, Hidden — and why moderation flagged it. The right side, Waiting for a decision, is the queue.

• Approve — shows a held or waiting item. Your decision overrides the AI's.
• Retry moderation — appears when items are held because the check could not run; runs it again for all of them.
• Hide — takes an item off every screen within seconds; Show again brings it back.
• Pin — keeps an item at the top of the screen. Several can be pinned at once — up to Pinned messages at once in the board's Settings (1–10, default 3). They sit in their own area above the stream, in the order you pinned them: stacked on a portrait TV, side by side on a landscape one. Pinning one more when the board is full unpins the oldest, and the panel tells you so. Unpin takes one off.
• Highlight — draws an item larger.
• Pause screen — the screens freeze on what they show; nothing new appears until you choose Resume screen, then what arrived meanwhile flows in.

Screen preview shows exactly what the TV shows: it reads the same answer as the screen link, so hidden, held, filtered and waiting items never appear in it, a paused screen stays paused, and the privacy filter applies. It also works before the event starts or while the board is switched off, so you can check the screen in advance.

Seeing the boards needs the See the live feed boards permission; settings, screen links and moderation need Run the live feed. Admins and supervisors have both by default; give them to anyone else in Settings → Team.

Who can open the screen

Each board has its own Who can open the screen setting:

• Signed-in team members only — the default for a new board. The screen link opens only for someone signed in to your workspace with permission to see the live feed. Without a session it asks them to sign in, then comes back to the screen. The link alone is not enough.
• Anyone with the link — the screen opens without signing in. Use this for a TV that cannot stay signed in. Anyone the link reaches can see the screen.
• No screen — no link works; you watch the board in Screen preview on the board's page.

Boards created before this setting existed keep Anyone with the link, so a screen already on a stand keeps working.

The screen link

In Settings → Screen link, choose Create link. Open the link on the TV's browser — full screen and read-only. You can copy it, scan its QR code with a phone, or open it from Open screen.

• Link expires — the drawer always says either Link expires: <date> or No expiry. Pick a date and choose Set expiry, or Remove expiry — the link itself stays the same, so a TV already showing it keeps working until then. Nothing expires unless you set it.
• New link — replaces the link; the old one stops working at once. Do this if a link was shared further than you meant.
• Turn off link — no link at all.

The screen shows only what your filters let through; names and personal data are removed on our side, before anything reaches the TV.

The board's own Starts / Ends dates are separate from the link: they decide when the board collects. With no dates set, it collects until you turn it off.

The screen updates within seconds and keeps working on unreliable Wi-Fi: when the connection drops it says Reconnecting… and catches up by itself.

Screen language (board settings) sets the language of the screen's own texts — title, Write to us labels, counters, the waiting and empty messages. Viewer's browser (auto) follows the TV's browser. Arabic, Hebrew, Persian and Urdu screens run right to left. Messages are always shown as they were written. Screen preview in the panel uses the same language. To try another language on one screen only, add ?lang=de (or another language code) to that link; it wins over the setting for that link.

The screen fits itself to the display: on a landscape TV the Write to us codes sit in a column on the right; on a portrait (vertical) screen the feed takes the full width and the codes become a row under the title. Text grows with the screen, so a 4K TV shows the same wall as a 1080p one, only sharper.

On the screen

• Write to us codes — QR codes for WhatsApp, Instagram, Messenger, Telegram, your web chat, a phone number or any link, each with a short label. Visitors scan them and write to you; what they write appears on the wall.
• Today's counters — messages per channel today, new conversations, and how fast Wiz answers (the median of today's replies).
• Image gallery — images that passed moderation, in a side strip.
• Theme — dark or light, and a title of your own at the top.

AI portraits on the screen

When you run Image Studio at an event, the portraits it makes can come up on the stand screen — full screen, a few seconds each. Only the generated portrait is ever shown, never the photo the visitor sent.

Turn it on in the board's Settings → AI portraits on screen:

• Show AI portraits full screen — off by default.
• Portrait profiles — tick the profiles whose portraits this screen shows. None ticked means every portrait profile.
• Timing — always shown, even before the spotlight is on:
  • How long each portrait stays (3–30 seconds, default 5): how long a portrait stays full screen before the wall comes back.
  • Wait between two new portraits (5–120 seconds, default 15): when several arrive at once, the next one starts this long after the previous one started.
  • Show earlier portraits again (on by default) and Show earlier portraits every (5–300 seconds, default 15): when nothing new arrives, an earlier portrait comes back this often. Turning repeating off greys out its interval.

What the visitor sees. Only while a board with this switch on is live, right after "your portrait is being prepared" the visitor is asked "Shall we show your portrait on the stand screen too?" with Yes, show it / No buttons (on WhatsApp, Instagram, Messenger and Telegram; elsewhere they type the answer). A typed answer can be a natural word in English or Turkish — göster, evet, olur, tamam, tabii, 👍, yes, sure, ok for yes; gösterme, hayır, istemiyorum, yok, no, don't for no — as long as it is the whole message. The lists are in the profile's texts (screen_consent_yes_words, screen_consent_no_words, comma-separated). Nothing is shown without a yes; no answer means no. A yes after the portrait was delivered still counts. Writing remove from screen (or ekrandan kaldır) at any time takes it off every screen. The wording is in the profile's texts (screen_consent_*, screen_revoke_keywords, screen_revoked) in AI tools.

What the screen does.

1. A new portrait pauses the wall (new messages wait) and comes up large for 5 seconds, then the wall carries on.
2. Several at once queue up: the next one starts 15 seconds after the previous one started.
3. When nothing new has come for 15 seconds, earlier portraits come back one by one, every 15 seconds — every portrait of the board's profiles, newest first, then round again. A new one always goes first.

The portrait goes through the board's AI moderation (when on) and Approve before showing (when on) like any other item. The moderation panel lists portraits under AI portraits, with Hide; In the spotlight marks the ones the screen may show right now. A portrait also leaves the screen when its file is deleted by the tool's retention setting or when the person's data is erased. With the privacy filter on, no name is shown next to the portrait.

Bonus code: one more portrait

Image Studio limits how many portraits one visitor can make a day. When a visitor has used them up, your stand team can give them one secret code, the same for everyone. When the visitor sends that code in the chat, they get an extra portrait.

Set it in the board's Settings → Bonus code:

• Give one more portrait with a secret code — off by default.
• Secret code — masked; Show the code reveals it, Copy the code copies it, Generate fills in a short readable one (like KRT-482). Capital letters, spaces and dashes do not matter when visitors type it. At least 5 letters or digits.
• Extra portraits per code (1–10, default 1) — how many more portraits the visitor can make in the next 24 hours.
• The code is valid … times per visitor (1–20, default 1) — leave it at 1 so nobody can use it again and again.
• How many times it has been used so far.

It applies to the profiles ticked under AI portraits on screen (none ticked = every portrait profile).

What the visitor sees. The daily limit message does not change and never mentions a code: only your team knows it exists. A visitor who sends exactly the code gets "Your code works — one more portrait is yours. Send a new photo whenever you are ready."; sending it again past the limit gets "You have already used this code." Both are in the profile's texts (bonus_granted, bonus_already_used) in AI tools. Wiz and flows do not answer the code. A wrong code is an ordinary message.

Never on a screen. A message that holds the code — alone, or among other words such as "my code is KRT-482 please" — never reaches the wall, the moderation queue, the screen preview or the screen link, and the code is never in what the screen reads. Messages that already carried the code before you set it are taken off every board when you save. Only the code alone is redeemed; a message that merely contains it is answered like any other message, but still never shown. Only members who can manage the board see the code.

Event set-up checklist

1. Create a board for the event. Pick its channels and, if you like, a hashtag. Set Starts and Ends to the event's dates.
2. Keep AI moderation and the privacy filter on. For a busy public space, turn on Approve before showing and keep someone on the queue.
3. Check that the AI is set up and the wallet has credit — without them every item is held.
4. Add Write to us codes for the channels you want visitors to use, and test each code with a phone.
5. Decide Who can open the screen: if the TV can stay signed in, keep Signed-in team members only and sign in on the TV; otherwise choose Anyone with the link. Create the screen link — with an expiry after the event if you want one. Open it on the TV, switch the browser to full screen, and turn off the TV's sleep timer.
6. Send yourself a test message on each channel and watch it appear. Try Hide and Pause screen once so the moderator knows them.
7. Showing AI portraits? Turn on AI portraits on screen, send yourself a photo through the portrait profile, answer Yes, show it and watch it come up.
8. After the event, turn the board off or let the link expire.