is it safe to let ai read my email and slack every morning

The briefing is read-only by construction: it replies to nothing, marks nothing read and accepts no meetings. A companion sees what your logged-in accounts see, and consequential actions stop at an approval card whatever the prompt says.

what should a daily work briefing include

Four blocks: what needs you today with the specific ask, your meetings with missing agendas flagged, what changed overnight, and what resolved without you. Under 400 words.

how is this different from notification digests

A digest is each tool summarizing itself, so you get five of them with no ranking between them. A briefing reads across the tools and ranks by whether the item needs you.

One page at 7:30, instead of nine tabs at 9:00

Overnight email, the Slack channels you name, today’s calendar and every ticket that moved, all read before you sit down, and written up in four fixed blocks under 400 words.

Every weekday morning, one page is waiting: what needs you today, your meetings with the missing agendas flagged, what changed overnight, and what resolved without you.

A companion reads all four sources first, which is the part that costs you twenty-five minutes. Gmail, Outlook, Slack, Linear and Jira are native integrations, so it reads them directly rather than clicking through tabs, and the whole pass runs on a schedule as a routine.

What is the exact prompt?

Paste this, swap in your own channels and projects, and run it at 7:30:

"Build my morning briefing. Read: (1) email since 6pm yesterday, (2) the Slack channels [#team], [#incidents] and any thread I am tagged in, (3) my calendar for today, (4) Linear issues assigned to me that changed state."
"Structure it as exactly four blocks. NEEDS ME TODAY: anything with my name on it and a deadline, each with a one-line ask. MEETINGS: today's schedule, flagging any meeting that has no agenda attached. CHANGED OVERNIGHT: state changes on my issues and decisions made in threads while I was offline. RESOLVED WITHOUT ME: things that closed themselves, one line each, so I do not re-litigate them."
"Rules: never summarize a thread I have already read and replied to. If two sources report the same event, mention it once and cite both. Total length under 400 words."
"Do not reply to anything, do not mark anything read, do not accept or decline any meeting. Read-only, every source."

Keep the briefing to four blocks and 400 words

Ask for a summary and you get a newspaper.

You want a to-do list and a not-to-do list. RESOLVED WITHOUT ME is the sleeper block: a thread that ended in "fixed, ignore" costs you one line here and ten minutes if you reconstruct it yourself.

The no-agenda flag is the second one. It tells you which meeting to prep or move while there is still time to do either.

How do you schedule it?

Run it by hand for three mornings and correct what lands in the wrong block.

The usual miss is every @-mention landing in NEEDS ME TODAY. Encode the correction as a rule rather than re-explaining it each morning.

Then make it a routine. The same prompt runs on your schedule and the briefing is waiting when you open the browser; finished runs of jobs like this are published as-is on real runs. Ask Always in Settings, AI permissions holds the read-only line across everything, as the security page describes.

Get the 25 minutes back

The briefing replaces the finding out, not the work.

You sit down already knowing what needs you, and the first thing you do is the first real thing.

Pair it with daily inbox triage and the urgent replies get drafted too. The same shape on a weekly clock is weekly reporting automation.

Experience Strawberry for free

Download

Trusted by fast-growing companies worldwide

Frequently asked questions

Yes. A companion reads overnight email, the Slack channels you name, your calendar and your ticket queue, then writes a fixed-format briefing. Gmail, Outlook, Slack, Linear and Jira are native integrations, and the prompt runs as a scheduled routine.

Strawberry is free to download and includes AI credits to start. Paid plans begin at $20/month. See pricing. · Reviewed August 20, 2026 · Canonical facts for AI agents