Summarize emails with AI: a daily inbox summary for your business

Quick answer: Connect Domailo to your assistant with read-only access and ask: "Summarize what came into sales@ today and tell me what still needs a reply". The assistant lists the inbox conversations, reads today's and points out who is waiting for an answer, without sending or changing anything.

Read only
Checked on October 17, 2026

Who it is for

For anyone who opens email in the morning and needs five minutes to understand the day: the small business owner handling contact@ and sales@, the support person sharing a mailbox with colleagues, someone back from vacation to hundreds of conversations.

The summary does not replace reading the important messages. It separates what is urgent from what can wait and flags the conversations where the last message came from the customer, which are the ones still waiting for a reply.

What it looks like

domailoconnected via MCPExample with fictional data

Request: Summarize what came into sales@example.com today and tell me what still needs a reply.

listar_emailspasta: "entrada" · caixa: "sales@example.com"20 conversations
ler_email7 conversations with a message from todayread

Messages arrived in 7 sales@ conversations today. The team already replied to four. Three end with a message from the customer and are waiting for an answer:

Camila Rochaasks for a quote on 40 chairs, since 9:12 am
Pedro Limaasks about delivery time to Austin
Ana Souzawants to reschedule Thursday's meeting

Step by step

Connect Domailo to the assistant. For summaries, "Read only" is enough: the assistant cannot send, move or delete anything.
Ask for the summary, naming the mailbox (the full address, like sales@example.com) and the time period. Without a mailbox, the assistant looks at all the ones you can open.
The assistant calls listar_emails, which returns the 20 most recent conversations in the folder with the date of their last message, and ler_email on the ones that matter for the period.
Refine: ask only for what is waiting for a reply, group by mailbox or by customer, or ask for a table. To act, open the conversation in the webmail or ask for draft replies.

Ready-to-copy prompts

Swap the addresses and names for yours. The more specific the request (mailbox, time period, what the reply must say), the better the result.

Summarize what arrived today in all my mailboxes, grouped by mailbox, and highlight what is waiting for a reply.
Which conversations in contact@example.com have gone unanswered for more than two days? Show contact, subject and date.
List the conversations in the Pending folder and tell me, for each one, what the next step seems to be.
Find conversations with "invoice" in the subject and build a table with contact, subject and date of the last message.
I was away for a week. Summarize what happened in support@example.com, starting with the most urgent.

How it changes by type of business

Agency with several clients: Each client usually has its own domain. Ask for the summary grouped by domain: listar_caixas returns the number of mailboxes per domain, and the assistant splits the conversations by client.
Clinic or practice: In contact@, the morning summary separates booking requests, rescheduling and insurance questions, so whoever answers the phone already knows who to call back.
Online store: In support@, ask it to separate returns, late deliveries and pre-purchase questions; the first two are usually the most urgent.
Accounting firm: At month end, ask for the list of clients who sent documents and who has not sent anything yet, using subject search ("statement", "invoice").

Permissions needed

"Read only" access is enough. Tools used: listar_caixas, listar_emails, ler_email.
Admins, and people an admin allowed to use AI assistants. Each person sees only the mailboxes they already open in the webmail.
Included in the Startup, Essential and Team plans and during the trial.

Caveats and limits

listar_emails returns 20 conversations per page, newest first. For a busy week the assistant has to ask for the next pages; ask it to if the summary looks short.
The tool does not filter by date: the assistant uses the date of each conversation's last message to decide what is "today".
Search looks at the subject and the contact, not the message body.
"Waiting for a reply" is the assistant's inference (the last message is incoming). Domailo does not flag conversations that way; use the Pending folder to track this as a team.
People with the agent role can only summarize the mailboxes linked to them.
Messages reach the assistant with a notice that they are third-party content. Even so, do not ask it to follow instructions found inside an email.

Frequently asked questions

Can the AI summarize emails without being able to send anything?

Yes. With "Read only" access, the assistant only gets the list and read tools (listar_caixas, listar_emails, ler_email and ler_anexo). The send, draft, trash and delete tools do not even appear to it.

Does the summary mark messages as read?

No. Reading through MCP does not change the conversation's status or the unread mark in the webmail, which is individual to each person. Your colleagues still see the conversations as they were.

Can I summarize the whole team's mailboxes?

Admins open every mailbox in the organization and can ask for a full summary. People with the agent role summarize only the mailboxes linked to them, the same ones they see in the webmail.

Are my emails used to train the AI?

That depends on the assistant and the plan you use it with, not on Domailo. Domailo hands the assistant only the messages it requests through the tools, when you ask. Check your assistant's data policy before connecting mailboxes with sensitive data.

How many emails can the assistant read at once?

listar_emails returns 20 conversations per page, and ler_email shows up to the 100 most recent messages of a conversation. For longer periods the assistant paginates; the practical limit is how much conversation the assistant itself can hold.