The problem this solves
The first hour of most marketing days disappears into tab cycling: the ad account, three competitor sites, an intent dashboard, each behind its own login and its own idea of what matters. By the time you have assembled a picture of what happened overnight, the morning is half gone and the picture is already aging. Teams usually patch this with a static digest email, and digests have a predictable fate: read closely for a week, skimmed for another, then filtered into a folder nobody opens.
The deeper problem is that briefings and conversations live in different places. The summary arrives in one surface, the discussion happens in Slack, and the context connecting them evaporates in transit. When someone asks whether today's ad anomaly is the same one from last Thursday, answering means leaving the conversation, re-running the analysis somewhere else, and returning with a screenshot. The thread has moved on before you get back, and the question quietly dies.
What a working team actually needs is a briefing that behaves like a colleague. It shows up once, in the channel where decisions get made, and then it stays in the room. Follow-ups get answered in place, grounded in the same evidence the briefing was built from, so the morning read turns into the morning conversation. The distance between noticing something and deciding what to do about it collapses to a single reply.
How the mission runs
- Set the cadence and the channel once. You tell the agent where the briefing lands - #general, in this playbook's prompt - and what counts as news for your business: which competitors, which ad accounts, which intent topics. That definition is stored with the mission, so every subsequent morning runs against your criteria without you restating them.
- The overnight sweep runs before you sit down. Ahead of your morning, the agent checks Ad Libraries for new competitor creative, scans Google Ads for spend and performance anomalies against your recent baseline, and pulls the accounts surging on your topics from Intent Data. Anything below the bar of genuinely notable gets filtered out rather than padded in.
- One thread posts to Slack. The briefing arrives as a single parent message with the two or three items that matter, each with a one-line reading of why it matters. No link dump, no wall of charts. It is written to be read in ninety seconds on a phone, with the evidence held behind it for whoever asks.
- Follow-ups are answered in the same thread. When you reply with a question - how does this compare to last week, which campaigns drove the anomaly, who else is surging - the agent answers in the thread with full memory of what it posted and the data underneath. Anyone in the channel can ask; the context is shared, so answers build on each other.
- The conversation carries across mornings. Thread memory persists between runs, so tomorrow's briefing knows what you flagged today. If you asked it to watch a specific competitor campaign, the next briefing reports on it unprompted. Credits for the sweep and every follow-up draw on the same workspace meter, so cost stays visible in one place.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
A daily Slack thread in your chosen channel: overnight competitor moves from Ad Libraries, anomalies from your Google Ads account, and the accounts surging on your topics from Intent Data, compressed to the items worth your attention. Under it, a running conversation where every follow-up gets a sourced answer, and a briefing tomorrow that remembers what you cared about today.
Make it yours
- Narrow the scope to a launch week: brief only on the two competitors that matter for the launch, and have the agent flag any new landing page or creative within hours instead of daily.
- Split the audience: a short version to #general for the whole team, with the full evidence version threaded to a private channel your leadership actually reads.
- Add a Friday edition that recaps the week's briefings, what was asked in the threads, and which flagged items turned into real changes worth carrying into next week.
Frequently asked questions
Does the agent actually remember earlier messages in the thread?
Yes. Thread memory is a product primitive, so a follow-up like "compare that to last Tuesday" resolves against what the agent posted and what was asked since. The same memory carries across days, which is what lets a Thursday question reference a Monday briefing without you re-explaining anything.
Will it take actions from the briefing thread, like pausing a campaign?
Only if you ask, and anything consequential runs through an approval gate. The briefing itself is read-and-report. If a follow-up turns into "pause that campaign", the agent states what it is about to do and the projected cost, and waits for your explicit yes in the thread before touching the account.
How is the cost of daily briefings plus follow-ups metered?
Everything draws on one workspace meter, whether the work started in the console or in Slack. The scheduled sweep has a predictable daily cost, follow-ups meter with the depth of the question, and you can set a threshold above which the agent asks before running an expensive dig.