Privacy Policy

Conjoiner is built by Aremesco LLC. This page says what happens to information you give us — on this website today, and inside the product when it opens.

Two things are true right now, and they shape everything below. This website does exactly one thing: it takes an email address for a waitlist. And Conjoiner itself is in private development — you cannot sign up for it yet, so the sections about the product describe what it will do rather than what it is doing with your data today.

The short version

• The only thing this site asks you for is an email address, and we use it once: to tell you when Conjoiner opens. Nothing else, nobody else.
• We never store your IP address. The rate limiter keeps a keyed hash of it instead, which cannot be turned back into an address.
• No advertising, no ad networks, no tracking cookies. We do not sell anything to anyone.
• Conjoiner will send you a text message for exactly one reason: a one-time code when you sign in. Never marketing. The details are here.
• Your content never trains a model. We measure how the product is used — counts, timings, failures — never the words in your documents. Training.
• Nobody at Aremesco can open your documents or chats: no support tool does it, because none exists. If someone here needs to see your work, your administrator invites them as a collaborator and revokes them the same way. Who can open them.
• We do not age your data out on a timer, and we do not pretend to. When you delete something it goes: out of live systems immediately, out of our backups within 35 days. Records already copied into a monthly archive leave when your organisation’s archive is expunged. Retention.

Who we are

Aremesco LLC is a Tennessee limited liability company. It is the company behind Conjoiner, and is responsible for this site and this policy.

That is the address to write to about anything on this page.

This website

The email address you give us

When you join the waitlist we store three things: the address, a short label for where the form was submitted from, and the time. Nothing else — the form asks for nothing else.

We use it once, to tell you when Conjoiner opens. We do not sell it, rent it, trade it, or hand it to anyone for their own purposes. If we ever want to use the list for something broader than that one message, we will ask you first rather than quietly widen this page.

The list lives in a hosted Postgres database. You can have your address removed at any time; see Your rights.

Your IP address, which we do not keep

The waitlist form is rate-limited to five submissions per hour from one network address, so a script cannot flood the list. Doing that needs a way to recognise repeat submissions, and the obvious way — storing the IP address — would turn a spam control into a log of who visited.

So we store a keyed HMAC-SHA256 of the address instead of the address. The key is a secret held outside the database. Without it the stored values cannot be matched back to any IP address, including by us, and the table is not a lookup table of visitors. The rows are pseudonymous and get cleared out periodically by hand; nothing schedules it.

Analytics

We count page views through a cookieless analytics service. Per its provider’s documentation, each recorded page view may carry: the time, the URL and route, the referring site, filtered query parameters, a coarse location (country, region, city), device type, operating system and browser version. Visitors are distinguished by a hash derived from the incoming request rather than by a cookie, and that identifier is discarded after 24 hours. What we see is aggregate counts. There is no ad network and no cross-site tracking.

Cookies

This site sets none. There is no cookie banner because there is nothing to consent to.

The providers behind all this

We run on other companies’ infrastructure, and each of them can see the data it handles in the course of running its service. By category, those are:

• cloud hosting, content delivery and infrastructure;
• database hosting;
• analytics, as described above;
• email delivery, for the mailbox you write to and the one message we send you;
• identity and sign-in, for the product;
• key management, for the product;
• SMS delivery, for the sign-in codes described below.

None of them has our permission to use your data for anything beyond providing that service to us.

Conjoiner, when it opens

Conjoiner is a collaborative documents product for teams, where people and AI agents edit the same page and every change is attributed. It is in private development. Nothing in this section is collecting anything from you today; it is here so that what we intend is written down before it happens, and because a phone carrier reviewing our right to send sign-in codes needs to read it.

Your account

Signing in is handled by a managed identity service, which holds the credential. Our own database stores a mapping to your identity and the profile fields you give us — an email address and a display name — not a password. If your employer uses Enterprise SSO, you sign in against their identity provider and we receive the identity attributes it sends us.

Text messages, and the one reason we send them

Conjoiner sends SMS for exactly one purpose: a one-time passcode used as a second factor when you sign in. We do not send marketing texts, promotional texts, or any other kind of message to your phone. There is no SMS list to be on, and no way to end up on one.

How you opt in. You add a phone number yourself, inside your own account, after you have signed in. On the Security page you choose to add a phone as a second factor, enter your number, and agree to receive one-time sign-in codes by text at that number. We then text a code to that number, and the number is enrolled only once you enter that code back. Nobody else can add a phone number to your account, and no number receives a message before that confirmation.

How often. One message each time you sign in using this factor, plus one when you first enrol the number. Frequency depends entirely on how often you sign in.

What a message looks like.

Conjoiner: 123456 is your sign-in code. It expires in 10 minutes. Do not share it. Reply STOP to opt out.

STOP and HELP. Reply STOP to any message to stop all texts from us to that number. Reply HELP for help, or reach us at support@conjoiner.ai. Message and data rates may apply. Carriers are not liable for delayed or undelivered messages.

Removing your number. Remove the phone on the same Security page and every message to it stops. You never have to give us a number at all: an authenticator app and a passkey are the other two second factors, and a phone number is never used to sign in or to identify you.

We do not share your phone number. Your mobile number and your consent to receive text messages are not sold, rented, or shared with anyone for their marketing or promotional purposes, and are not shared with any third party except the messaging provider and carrier needed to deliver the message itself. Messages are sent from the toll-free number +1 844 942 3521 and can be delivered to United States numbers only.

These rules are the program. No sign-in codes are being sent yet, because Conjoiner is not open yet; when they are sent, this is what happens.

What you create

Documents, comments, chat messages, the history of every change, and a record of each action taken by a person or an agent. That last one is a feature and not a by-product: attribution and an audit trail are the reason it is safe to let an agent edit your document.

AI providers and your keys

Conjoiner does not resell inference. You bring your own metered API key for the AI provider you choose, or run a model on your own hardware. A key you give us is stored encrypted with per-tenant envelope encryption in a managed key-management service, is never logged, and is decrypted only in the moment a request is made.

What you send to a provider is then in that provider’s hands and is governed by your agreement with them, not by this page. Conjoiner will never route requests through a flat-rate consumer subscription — only a metered API account — and will never ask you to sign in to a vendor’s chatbot.

Training, and what we do measure

Your content never trains or fine-tunes any model — ours, or anyone else’s. Not your documents, not your chats, not your comments, not your edit history.

Usage is measured, and that is a different thing.We record counts, timings and failures: how many requests an agent made, how long they took, how many tokens they cost, what went wrong. That is what makes an agent’s spend visible to you rather than a surprise on a bill. It never includes the words in your documents.

Who at Aremesco can open your documents

Nobody. There is no product feature, console or support tool through which anyone at Aremesco can read your documents or your chats. Not a gated one, not an audited-but-permitted one, not one that asks your permission first. This is not a policy we have written against using such a path — there is no path to use.

So when you want help with something that needs a person here to see your content, that person gets in the way anyone else does: an administrator at your organisation invites them as an external collaborator, scoped to what that account can see and revocable by your organisation at any moment. Nobody at Aremesco is special-cased anywhere in the product. An invitation you can withdraw is a different thing from a vendor privilege you cannot.

The boundary, stated rather than stretched. That is a claim about the product, and we will not inflate it into one we cannot keep. Running a service means the people who operate ours hold infrastructure access — to databases, to logs, to object storage — as they do at every company that runs software. What does not exist is a way to turn that into reading your work inside Conjoiner: no feature, no button, no support mode. Access to your content comes from an invitation your organisation issued, or it does not happen.

Where it lives

Product infrastructure runs in a single region inside the United States, and nowhere else. That is not a preference: a policy on our cloud accounts denies every action outside that region, so a service cannot be stood up elsewhere by accident.

How long we keep things

Most privacy policies have a sentence here promising that everything is erased after some number of days. Ours does not, because it would be false. Nothing here ages out on a timer. What follows instead is what actually happens.

• Your waitlist address stays until Conjoiner opens and we have told you, or until you ask us to remove it.
• The pseudonymous rate-limit rows are cleared out by hand from time to time. They contain no address and no identifier.
• Analytics are retained by the provider under their terms; the 24-hour visitor hash is theirs to discard and they say they do.
• Documents you deletein Conjoiner go to trash first, and are permanently destroyed once your organisation’s trash window has passed — 30 days unless your administrators set it differently. That window is a grace period for changing your mind, and its clock starts when you press delete. That destruction is real: content objects and rows are removed, not flagged.
• Deleted content leaves live systems immediately and our backups within 35 days. This is the second, separate clock, and it starts at the moment content is actually destroyed — when you empty the trash, when the trash window above runs out, or when you delete something that never went to trash at all. It applies to an account you close as much as to a document.
• The monthly archive is the exception, and we would rather name it than let you find it. Chat messages and agent-run records from months that have already closed are copied into an archive, one object per organisation per month. A single document’s rows cannot be lifted back out of one: the object is your organisation’s whole month, written once, and editing it would destroy the very property that makes it worth keeping. So a document you delete leaves the live systems and the backups on the clocks above, while that copy stays until your organisation’s archive is expunged — an action an administrator of your organisation can take, all-or-nothing, for the whole archive. There is no timer on it and we are not going to invent one. In the meantime nothing in the product reads those objects: no page, no search, no feature.
• The audit trail is kept indefinitely, on purpose. It is the record that makes every other deletion safe — when a document is destroyed, the fact that it was destroyed, and by whom, survives in the audit log. A trail with a hole in it is not a trail, so we do not put holes in it. It holds the fact, never the content.

Honest and unusual beats familiar and wrong.

Your rights

These are yours wherever you live. We do not ask which law covers you before honouring them, and we do not charge for them:

• Tell you what we hold about you.
• Give you a copy of it, in a form you can take elsewhere.
• Correct it.
• Delete it — including taking your address off the waitlist.

Email hello@conjoiner.ai. Deleted content leaves live systems immediately and our backups within 35 days. Records already copied into a monthly archive leave when your organisation’s archive is expunged, which an administrator of your organisation can ask for at any time; there is no schedule on that one, and Retention explains why.

For organisations. If your team needs a data processing agreement, ask us and we will provide one. Where the data includes personal data from the European Economic Area or the United Kingdom, standard contractual clauses are available as part of it.

Age

Conjoiner is a product for people at work. You must be 18, or old enough to enter a binding contract where you live, to use this site or to join the waitlist. We do not verify anyone’s age — this is a term, not a gate, and we would rather say so than imply a check we do not perform. We do not knowingly collect anything from a child; if you believe we have, email us and we will delete it.

Changes to this policy

The date at the top of this page is the date it last changed. If a change makes a material difference to what happens to your data, we will say so here before it takes effect — particularly the commitments this page makes plainly: the single use of your email address, the promise that your content never trains a model, the absence of any staff path into your documents, and the deletion timeline.

About this document

It was written by the people who built the system it describes, in plain language, and every claim in it is meant to be checkable against something real. It is not legal advice. It has not yet been reviewed by counsel, and it will be before Conjoiner opens. If you spot something here that does not match what we actually do, tell us — that is a bug, and we will fix it.

See also our Terms of Service.