Privacy Policy

Just Say It is operated by ELAV8 AI LLC (“we”), 1343 Main Street, Suite 205, Sarasota, FL 34236, United States. This policy covers the website at https://justsayit.ai and the product at https://justsayit.ai/app. Questions go to [email protected].

1. What we collect

Identifiers and account information

When you sign in with an ELAV8 account: an account identifier, the name and picture the sign-in provides, and your e-mail address. The address is kept in your own browser’s encrypted cookie and in the server’s memory while you are signed in — never in a recording or a log — and is used for one thing: the mail the product itself sends you (a welcome the first time you sign in, a warning when a shared pool of minutes is nearly out, a receipt when minutes are added to your month, the notes you ask for, and a meeting’s notes after it ends — the “Meeting notes by e-mail” row). Where the product must recognise your address again later — an invitation into a team sent to it, a calendar whose meetings Ziggy joins, a share made for it — it keeps a one-way code of the address in a file, from which the address cannot be read back, never the address itself; the e-mail delivery service receives the address for each delivery, and the one place it is written in plain text is the e-mail you receive. ELAV8 holds your plan and your payment details; we receive your plan and the minutes it allows.

Enterprise enquiries

What you type into the form on the Enterprise page — the company, its size, its region, what you need and an address to answer — is kept in a file the makers read, sent to us as one e-mail and to you as an acknowledgement, and used to answer you. Nothing else is done with it.

The creator program

If you apply or register a post: what you type into its forms — your name, links to your profiles and to your videos, your niche, where your audience is, your address, and for a post the account that posted it, your code and the day it went up — kept in files the makers read, sent to us as one e-mail and to you as an acknowledgement, and used to run the program: approval, checking views, and paying you. A payout service holds your payout and tax details; we never see your bank details.

A creator’s link

If you arrived through a creator’s link, that creator’s code is sent with a trial or a sign-in, and we count how many came through each creator. The count names the creator; it is not tied to you, your account or your recordings.

Audio and its record

While a session is live, the sound from your microphone streams to a speech service and is transcribed as it goes; the transcript, the slides the conversation composes and the notes — with the session’s events and the names of the people in the room — are the recording. No sound file of a conversation is kept: the microphone’s stream is transcribed and discarded. The one sound we keep is a voice sample you record yourself for a film (the “Faces, photos and voices” row), and only for as long as that row says.

Room participation

In a room: each participant’s name as given, their account identifier if signed in, when they joined and left, and the type of browser their tab reported. Cameras are not recorded, except as a team photo you took yourself or a snapshot the host kept with your face (the “Faces, photos and voices” row). A screen share is not recorded; when the person sharing lets the room hear its sound, the words it plays are transcribed with the session, marked as the shared screen’s, never as a person’s, and never put on the slide — the person sharing is asked once, before the first sound is heard, to confirm they have told the people whose words it carries. A call the host adds from another browser tab is transcribed into the session in the same way, marked as the call’s; the host confirms, before it starts, that the people on that call will be told.

Connected accounts

If you connect accounts or services of your own — a mailbox, a workspace or issue tracker from our list, any server you add by its address, or a tool running on your own computer through its link — they are used only when you ask: a question you say or type (“have I gotten anything about the offsite?”) becomes a search of them, or, on a server you added yourself, a call to a read-only tool you switched on. The best matches are read to write your answer and then discarded: neither the results nor the answer is kept in a recording, a file or a log — only counts, such as how many results came back (a question you say aloud is part of the conversation like anything else you say). Alone, the answer is spoken to you and waits in a thread only you can see; while others are in the room it appears only on your own screen, and nothing reaches the shared slide unless your setting for that connection, or your click, says so. When you show it, the slide carries a page of results — a title, who and when, a line each — under a tab of its own; shown while you are alone it is kept out of the recording, and shared with others in the room it becomes part of that meeting’s record. Signing a service in gives us a token that lets us read it on your behalf; it is kept encrypted, used for nothing but your own questions, revoked where the provider allows and deleted when you disconnect.

Faces, photos and voices

Pictures of people reach us in four ways, each by a choice. (1) Your own photo, which you give in Settings so that films and memes can star you, and a voice sample you record yourself so a film can speak your lines in your voice; (2) a team photo — when someone in a room starts one, a countdown shows on every screen and each person decides on their own screen, with a “Yes” under their own picture, whether their face is kept for that team’s memes and slides; (3) a snapshot — a host who has turned the switch on in Settings → Team faces, and accepted there that they are responsible for having permission from everyone whose face they add, has their tab take up to three frames of each other person’s camera at random moments during the meeting; a snapshot is held, usable nowhere, until the host confirms it; (4) a photo or a voice the host adds to their cast for a film, stating as they add it that they are that person or have that person’s permission. Each of these is a likeness — biometric information under some laws — and we treat it so: a vision model looks at a snapshot or a team photo only to check that it shows one clear face, and identifies nobody; a face is used for one thing, to picture that person — as the picture on their tile, in a meme or a slide picture made about what they said, or in a film the host makes — and never to recognise, match or find anyone; it is kept for three years after it was last used and then deleted automatically, or at once when the person removes it (Settings → Team faces → Remove, or Your photo → Delete), which deletes the files here and in our storage, and the films and pictures made from them; nobody receives it but the picture-making service named in section 4, for the one picture it is making. A snapshot is never taken of the host’s own seat, nor of a person whose connection comes from a place with its own law on consent to a likeness (Illinois, Texas and Washington today), nor when a person’s place cannot be told: such a person’s face is kept only by their own “Yes” on their own screen. The Data Retention page sets out the periods and how each is destroyed.

Trial usage

A session without an account is counted per network connection for one day, to limit trials; the address itself is not kept.

Tips by e-mail, and your choice about them

Your ELAV8 account may send you an occasional tip by e-mail about a feature of this product you have not tried — at most one a week, chosen from the record of what your account has used (above), never from what was said in a conversation. Before the first one, an account that was never asked sees a small card in the product with the exact words it will be recorded as having seen — “Tips by email: On · Off” — and your answer is kept by your ELAV8 account with those words and their version; we keep only that you answered, and which way, as a code. The product may also show you one such tip as a card inside the app; we record that it was shown, tried or put away — the tip’s name, never its words. Every tip e-mail carries a one-click unsubscribe your mail program can use without signing in, and the three switches live on your ELAV8 account’s e-mail settings page. When you press a tip e-mail’s button, the product opens on that feature with one line about how to try it; it counts that a tip e-mail was opened (the feature’s name, nothing about you beyond your measuring choice above) and tells your ELAV8 account which tip it was, so it can stop suggesting a feature once you have tried it. Nothing of this is sent while you are on a trial, not signed in, or a learner without an account of your own. Tips are switched on one at a time: until one is switched on it is only planned and recorded, and nothing is sent.

Ziggy in a meeting held elsewhere

If you send Ziggy into a meeting held elsewhere, or invite [email protected] to one, a meeting-bot service joins that meeting as a participant for you, shows your live slide as its camera, and carries the meeting’s sound to us while it is there; that sound is transcribed into your conversation’s record like your own microphone’s. The service posts a notice in the meeting’s chat as it joins and Ziggy says one line; it keeps no recording of the meeting (its storage is off), and it leaves when the meeting ends, when everyone else has left, when you say so, or after three hours. The people in that meeting are not our users: we keep nothing about them but the words they say in your record, and you are the one responsible for telling them.

Meeting notes by e-mail

A few minutes after a meeting ends — two after the host ends it for everyone, ten after the last person ends or leaves — each person who was there and is signed in gets one e-mail with what the record holds: the summary, what was decided, the action items and who owns them, the open questions, that person’s own items, and links to the slides, the replay and a download. Never the transcript. The links in that e-mail are a share made in the host’s name for that recipient: anyone who holds the link can open the slides for 30 days — a forwarded e-mail opens them too — and someone who was in the meeting is taken straight to its replay, while anyone else may keep the meeting in their own conversations after signing in; the host can revoke every such link earlier from the Share panel. If the host’s team keeps a picture of your face from that meeting, the e-mail says so and carries two links of your own — Keep my face · Remove my face — that act without a sign-in and work for 30 days. The host chooses who gets them — everyone who was there (the default), only themselves, or nobody — in Settings → Meeting notes; each person can stop the notes of others’ meetings there or with the one-click unsubscribe every one of these e-mails carries, which works without signing in. We keep, in the recording, which accounts it went to as one-way keys — never an address — so nobody gets the same meeting twice. A trial and a re-run get none. A guest who joined by name, without an account, is offered the notes on their own screen — when they ask for them, and when the meeting ends where the host sends the notes to everyone — for a free account: once they create one or sign in in that tab, their seat in the meeting becomes their account’s (the minutes they spent as a guest are not counted to it), the meeting appears in their conversations, and its notes go to the address their sign-in verified.

Usage and technical data

Per session, the minutes used and the volume of processing, to meter plans and run the service; server logs of codes and counts. How the site and the product are used, measured by us alone: the pages you open, the site that sent you, every control you press — recorded by a fixed name for that control (“the mute button”, “the invite link”), never by the words on it or anything you typed — which panels you open and for how long, which keyboard shortcuts you use, which refusals or errors you were shown (as codes), and the shape of a session — how long it ran, how many people took part, what kinds of things appeared on its slide — never what was said, written or shown. We also record which product features you used and when — a lesson, a quiz, a game, the assistant answering you, a picture or clip you asked for, the notes, a share, a film — as the outcome alone, never its content, and credited to the person who did it (the host or a guest), never to a learner. These records carry a random identifier, not your name or address; when you are signed in, that identifier is linked to your account. When you are signed in, we also tell your ELAV8 account — the identity and billing service you sign in with — which of these features your account used, when, and whether you hosted the room or joined it: the feature’s name and the moment, sent in numbered batches, nothing else — never what was said, written or shown, never anyone but the signed-in account holder who did it (a learner without an account of their own is never named), and nothing at all while you are on a trial or not signed in — so that one place keeps what each person has used across ELAV8’s products. That record is part of running the service and does not stop with the browser switch below, which stops what your browser sends. We use this to see what is used and what is not, and — once that step is built and this page says so — to choose which product tips, if any, to send you; you can turn the measuring off for your browser in Settings and on Your Privacy Choices. In the EEA, the UK and Switzerland the browser sends nothing until you have chosen on a small card; the choice is kept 13 months and asked again after. The site’s pages make no third-party request: the typeface is served from this site, and there are no advertising trackers.

Some of this is sensitive personal information under the laws of several states and countries, and we say so plainly: a face kept as a team face or a cast photo and a voice sample recorded for a film (biometric information); the contents of a connected account, read for the moment of your question; the access tokens that connect those accounts; and whatever the people in a room choose to say, which the transcript keeps as they said it. We collect each only by the choice described above, use it only for the purpose described there, and never sell personal information or share it for advertising.

2. Recording and consent

Every session is recorded in the sense above: its transcript, its slides and its notes are kept for the account that started it. In a room, the words of everyone present are transcribed and kept the same way.

The person who starts a session (the host) is responsible for telling everyone in the room that the conversation is transcribed and kept, and for obtaining their consent before they speak. In a number of jurisdictions — California among them — recording a conversation requires the consent of everyone taking part; the host must know and follow the law that applies to the people in their room. The Terms of Service make this the host’s undertaking.

The product tells people itself, on screen. The card a person sees before joining a room says, above the button, that the meeting is transcribed by an AI co-presenter and its slides and notes are kept for the host, that a face may be kept only by that person’s own choice or by the host under the host’s responsibility, and that joining means they agree; the wording they agreed to is recorded with their entry. Every screen in the room shows a “Transcribing” indicator whenever anyone is on the air — the host’s screen, a muted host’s, a guest’s who never spoke. And a guest who is in a room while it is transcribed sees a notice on their own screen saying so. In a room nothing is said aloud: the people are told on their screens, and by the host.

A person who does not wish to be recorded should not join the room, or should ask the host to end the session. A participant may ask the host to delete a recording; the host can do so from their history at any time.

Two more ways words reach a session carry no card of their own, and the host’s duty covers them. A host may add a call running in another browser tab, so that the people on that call are transcribed into the session too (marked as the call’s, never as a person in the room); before it starts, the host confirms on their own screen that those people will be told and that any consent the law requires has been obtained, and that confirmation is recorded with the session. A person sharing their screen may let the room hear its sound; before the first sound is heard they confirm the same for anyone whose words it carries. In both cases nothing on our screens reaches the people on the other end: telling them is the host’s undertaking under the Terms of Service, and a recording made without it is theirs to delete.

The same holds when you send Ziggy into a meeting held elsewhere, or invite [email protected] to one: the people in that meeting see the notice it posts in that meeting’s chat as it joins and hear it say one sentence when it first speaks there, and anyone in the meeting may remove it; nobody there is shown our screens, the recording and the notice of this policy, so you remain the one responsible for their consent, and the meeting’s words become part of your record, which you can delete.

3. Why we collect it

• To provide the service: transcribing you, composing the slide, answering you, keeping your notes, bringing back what earlier conversations you took part in decided, and metering your plan.
• To keep the service secure and reliable.
• To understand how the product is used and make it better — counted, never read.
• To tell you, kindly and at most once a week, about a feature you have not tried — by a card in the product, or by e-mail only once you have had the chance to say no (and never after you say it).
• To answer you when you write to us.
• To improve the product’s own judgement: we may use the records of conversations — the decisions the product made about them and the text they were made from, never audio, never a face or a voice — to tune the models we run ourselves for this product. Such a record is never sold and never handed to anyone else to train on, and a conversation you delete is gone from that material too.
• Not to sell or share personal information.

For people in the EEA, the United Kingdom and Switzerland, our lawful bases are these. Providing the service you asked for — the transcript, the slides, the notes, the memory, the metering — is the performance of our contract with you. Your consent is the basis for your own face or voice, for the measuring your browser does, and for tips by e-mail, and you may withdraw each where you gave it. Our legitimate interests — keeping the service secure, understanding how it is used in counts, and tuning our own models as described — are the basis for the rest, and you may object to any of them by writing to us. Where a host keeps another person’s face or adds their photo or voice, the host decides to do so and is responsible for having the right to; we process it on the host’s instruction, as the Terms say.

4. Who handles it for us

Service providers process personal information on our instructions and for our purposes only, in these categories:

• Cloud hosting and storage providers, in the United States.
• Speech and language processing services, which process your audio and transcript to provide the service, and your question and the matching parts of your connected accounts when you ask to search them.
• Identity and payment services (your ELAV8 account), which hold your account, plan and payment details, and which of the product’s features your account has used and when; we never see your card number.
• Network, security and media-relay providers, which carry the site’s traffic and the calls between people in a room.
• Public map and picture sources, which receive the place or the name you said and nothing about who said it.
• A web search service, which receives the words on the slide you asked about and your question, and nothing about who asked.
• A picture- and video-making service, in the United States, which receives a scene we wrote from the conversation and — only for a person whose face is kept as described in section 1: their own photo, a team photo they said yes to, a snapshot the host kept, or a cast photo and voice the host added — that person’s photo, and for a film their voice sample, to make the memes’ pictures and short video clips, the slide pictures, the films and the camera backgrounds; it receives no name, and the link it fetches a photo or a voice from expires within minutes.
• A vision model, through the speech and language provider above, which looks at a snapshot or a team photo only to answer whether it shows one clear face; it is told nobody’s name.
• An e-mail delivery service, in the United States, which carries the e-mails we send — the meeting notes, a share, a welcome, a receipt, the creator program’s mail — and so receives the address and the e-mail’s words for delivery.
• A meeting-bot service, in the United States, which joins a Meet, Teams or Zoom meeting you send Ziggy to or invite it to, shows your slide there and carries the meeting’s sound to us while it is in it; it keeps no recording of the meeting.
• An error-reporting service, which receives a report when the product itself fails — the kind of failure, where in our own code it happened, which version was running and the type of browser or server — never what was said, written or shown, never your name, address or account, and never your network address (a report from your browser passes through our own server first).
• For creators in the creator program, a payout service, which holds their identity, payout and tax details and pays them.

5. How long we keep it

• Recordings: until you delete them from your history — one at a time, or every conversation you hosted at once (“Delete all my conversations…” at the end of the list). A deleted recording is removed from our storage too, not only from your list, and the films, pictures and clips made from it go with it — here and in our storage; a copy a browser or the network cache fetched before the deletion may remain until it expires. A conversation you mark private is never brought back by the co-presenter and never shown in search.
• A trial’s recording: one hour, unless you sign in within it, in which case it becomes yours.
• Faces, photos and voice samples: three years after the last use, or at once when the person removes them; a face or a voice removed takes the films, memes and slide pictures made from it with it.
• Account information: while the account is in use, and until you ask us to remove it.
• Trial counts: one day. A waitlist address: until you ask us to remove it.
• Usage records: kept to understand the product over time. The link between them and your account is removed when you ask us to delete your data, or after 25 months in which it is not used — after that they cannot be traced to you. No network address is kept with them.
• A creator application or a registered post: while the program needs it, and as long as the law requires for a payment made; ask us to delete it sooner and we will, unless a payment keeps it.

The Data Retention page sets this out as one table — each kind of data, where it lives, how long it stays and how to delete it.

6. Memory across conversations

The co-presenter can answer “what did we decide” from earlier conversations — and only from ones that every person listening took part in, as the meeting’s own record of who was there says. When someone in the room was not in that earlier meeting, the answer goes to the asker’s own screen alone, and the co-presenter tells the room only that it did; when the asker was not in it, there is no answer. A conversation you mark private is never brought back and never shown in search. Asked to be exact, it may quote a line of an earlier transcript, with who said it and when.

7. Cookies and storage

Signing in sets cookies that keep you signed in; a trial sets one that lets its recording be claimed; your browser’s own storage keeps your preferences, the random identifier the site measures its own use by, and, for as long as the tab is open, the code of a creator’s link you arrived by. There are no advertising cookies and no cookies of ours set by anyone else; the one cookie not ours is a security cookie our network provider may set, and the human check on the trial door is served from that provider. The Cookie Policy lists them all.

8. Your rights

• Delete any recording from your history yourself, at any time — or every one you hosted, at once. Mark any conversation private.
• Ask us to tell you what we hold under your account, to correct it, to give you a copy, or to delete all of it: write to [email protected] from the address you sign in with. We confirm when it is done, and we never treat you differently for asking.
• If a security incident affects your personal information, we tell you at the address your account uses, without undue delay and as the law where you live requires.
• You may complain to the data-protection authority where you live; we would rather hear from you first.
• The Your Privacy Choices page says how these rights apply under the laws of California and other U.S. states.

9. Children

Just Say It is for people 13 and over; an account is for adults — 18 and over — who may hold an ELAV8 account. A child under 13 may take part only when a parent or teacher is the person who runs the meeting — a classroom, a family at home — present and responsible for them: what the child says is part of that adult’s recording, kept for the adult and deleted when the adult deletes it. We do not knowingly collect personal information from a child under 13 for ourselves, and the Terms forbid adding a child’s photo or voice to a cast; a host who keeps a team face of a child does so under the host’s own responsibility for the consent the law requires. A parent or teacher who believes a child’s information has reached us otherwise should write to [email protected] and we remove it.

10. Changes and contact

When this policy changes, the date at its foot changes and a material change is announced in the product. Write to [email protected] or to ELAV8 AI LLC, 1343 Main Street, Suite 205, Sarasota, FL 34236, United States.