Terms of Service

The service

reply2mail is an AI email agent that reads, classifies and (in your chosen mode) replies to your email. You are responsible for the email account you connect and for reviewing the agent’s actions, especially in auto mode.

reply2mail also lets you connect outside AI products (such as Claude, ChatGPT or Cursor) to your reply2mail account through our MCP server and API, using an OAuth connection or an API key. This feature ("Integrations") is new; we may change how it works, which products can connect and which actions are offered.

Your responsibilities

Only connect accounts you are authorized to use. You are responsible for messages sent on your behalf, whether by reply2mail’s own agent or by a connected AI product, once you have approved them or let them run without approval. We recommend starting in suggest mode and reviewing drafts before enabling auto mode.

Do not use reply2mail for spam, harassment, or any unlawful purpose.

Connected AI agents

You choose which AI products to connect and which permissions to give them, and you are responsible for that choice. A product you connect acts on your instructions and under its own provider’s terms; reply2mail does not control it and is not responsible for what it does with data it has read, including anything it sends, stores or shares through its own tools.

You are responsible for all activity carried out with your API keys and connections, including approval requests you approve and actions that run because you switched approval off. Keep API keys secret, do not put them in shared code or public places, and revoke a key immediately if you think it has leaked.

Only connect a product to mail accounts you are entitled to give it access to. Emails you receive may contain personal data of other people; you are responsible for having a lawful basis to let a connected product process it.

A connected product can read every email in the connected accounts, including password-reset and login-code emails from other services. The only exception is a connection made from ChatGPT, from whose results reply2mail leaves out emails it classifies as login codes, password resets or security codes; that classification can miss an email. Emails written by others can contain instructions aimed at the AI. Only connect products you trust with your mailbox.

Approvals

When a connected product asks to send a new email, reply, forward, archive, move to Trash, move, label, or act on several emails at once, reply2mail does not do it. It creates an approval request that you review in the reply2mail dashboard. Nothing happens unless you approve it there. Approval links or buttons inside an AI product or an email are not approvals; reply2mail only accepts approvals given in its own dashboard.

This is the default. On a paid plan you can switch approval off for one key or connected product and one permission group, in the reply2mail dashboard only. Its requests in that group then run without an approval request, within the limits shown when you switch it off; you instruct reply2mail to carry them out and are responsible for them as if you had approved each one. To let a key or product send without approval you confirm with your two-factor code or your account password. With the default settings, a send to an address the sending account has not written to before, or above the send limit, still becomes an approval request; a send held by the outbound safety filter always does; and irreversible actions (such as deleting a folder or cancelling an event) always require approval. What ran without approval is listed in History on the Approvals page and you are notified afterwards; sent emails cannot be recalled. You can return everything to approval with one switch, and approval returns automatically if you move to the free plan, change your password or turn off two-factor authentication. This setting does not apply to reply2mail’s own assistant or automatic replies.

Marking emails read or unread and flagging them run immediately, without an approval request.

When you approve a request, you instruct reply2mail to carry out exactly the action shown on the request, from the account shown, and you are responsible for it as if you had done it yourself. For bulk requests, the action applies to the list of emails shown when the request was made, not to emails that arrive later. Forwards do not include attachments.

Requests expire after 24 hours if you do not act on them. A request is cancelled if its target changes, if you revoke or pause the connection, or if you switch off the permission it needs.

Sent emails cannot be recalled. If reply2mail cannot confirm whether an approved action completed (for example because the mail server did not answer), it reports that the action may or may not have happened and does not retry it; check your Sent folder before trying again.

The outbound safety filter can hold an approved send that appears to contain secrets such as passwords or ID numbers and ask you to confirm again. The filter does not judge whether an email is appropriate; that review is yours.

Plans & billing

A free plan with a monthly mail quota is available. Paid plans (Starter, Pro, Business) add higher quotas and capabilities, billed via Paddle. You can cancel anytime; access continues until the end of the paid period.

Integrations is available on the free and paid plans, with limits that depend on your plan: how many keys and connected products you can have, how many calls a connection can make per minute and per day, how many approval requests can be created and waiting at once, and how many emails one bulk request can cover. The current limits are shown in the dashboard. We may change them; we will not lower the limits of a paid plan during a billing period you have already paid for, except to stop abuse or protect the service.

Calls from connected products do not use your monthly mail credits. Actions you approve run as normal mail actions.

If you move from a paid plan to the free plan, connections beyond the free plan’s limit are paused until you remove some or upgrade again, and permission groups you set to run without approval return to approval.

Suspension of Integrations

We may pause or revoke an API key or connection, switch off Integrations for your account, or suspend your account, with or without notice, if we reasonably believe it is being used: to send spam or unsolicited bulk email; to harass, defraud or impersonate anyone; to get around rate or plan limits (for example by creating several accounts or sharing keys between people); to attack, probe or overload reply2mail or a mail provider; in breach of a mail provider’s or the connected AI product’s terms; or in any unlawful way. We may also do this when a key appears to have leaked or a connection behaves in a way that puts your mailbox at risk.

We may switch off Integrations for all users at once to deal with a security incident. Requests waiting for approval then expire without being carried out.

Where we can, we tell you what happened and how to restore access. You can always revoke your own keys and connections in the dashboard.

Availability & liability

We work to keep the service reliable but provide it “as is”, without warranty. We are not liable for missed, delayed or incorrect emails, or for actions taken by the agent that you did not review. Your use is at your own risk.

Integrations is provided “as is”, like the rest of the service. Outside AI products can change their behaviour, misread emails, act on instructions contained in emails, or report an action as done when it is still waiting for approval. We are not liable for actions or statements of a connected product, for data it obtained through a connection you created, or for actions you approved or let run without approval.

Changes & contact

We may update these terms; material changes will be noted here. Questions: [email protected].