Your monitoring already tells you what happened. The AI morning brief is how comms leads now get to the decision faster, before standup.
I am the COO of Determ. The morning I am about to walk through is not mine: it belongs to the comms leads who have been on the MCP beta since we shipped it this spring. Mid-senior, in-house communications managers at fintechs, telcos, and regional energy companies, the people who spend the first 25 minutes of every day finding out whether the world moved overnight. I run a version of it on our own coverage too, which is how I know it holds up.
Here is what their 7 a.m. used to look like, and what it looks like now. The scenario is composite, stitched from how five teams describe the same workflow. The data shape and the conversation are real. Company and outlet details are illustrative.
Picture it. 7:02 on a Tuesday. The Determ digest is already in the inbox: overnight coverage, filtered to the topics the team tracks, grouped, summarized, and sentiment-scored. Around 180 mentions across 11 markets, the noise already stripped out. 58 minutes to standup. The tool has done its job. The open question is the one a digest cannot answer on its own: which of those 180 actually changes what gets done today?
What “the morning scan” used to mean
Until last year, even with the digest in hand, the morning was a judgment exercise done by hand. Read the digest. Open Determ to dig into whatever looked heavy. Check whether an overnight spike is still climbing or already fading. Glance at the comms channel in Teams for what was flagged, and the CRM for which of these reporters the team already knows. Then build a mental priority list while the coffee brews.
The digest said what happened. It did not say what to amplify first, or hold this morning against last quarter, or notice that one of the reporters sharing the story has never covered the brand before. That part was theirs. On a quiet day it took 20 minutes. On a busy one the digest ran long, standup started before the list was done, and now and then the thing that mattered was item 14, not item 1.
The pattern: the tool filtered the inputs, the judgment did not scale. Determ surfaced the coverage. The comms lead still ranked it, in their head, against context the digest did not carry.
What changed
This spring we shipped the Determ MCP. Technically, it connects Determ to AI assistants (Claude, ChatGPT, Cursor, and others) over a shared standard called the Model Context Protocol. Practically, it puts a team’s media data next to the assistant they already use to read email and draft documents. Determ already tracks the firehose: the same engine recently logged 16,000 mentions of a single launch in one day. The MCP does not change what we measure. It changes how you ask.
The morning now looks like this. Open the assistant. Type one question. The answer comes back in 30 to 90 seconds, sorted by what to do about it.
Here is a real exchange, lightly anonymized. The data shape is real; the specifics are illustrative.
7:02, the question
Anything from overnight worth acting on before the 9 a.m. standup? Weight by source authority and reach concentration, not raw volume.
7:02, the answer (Determ connected)
Last 14 hours: 180+ mentions across 11 markets, sentiment strongly positive. Three tiers.Amplify now. The feature you announced yesterday is getting picked up by a major industry outlet, and two reporters who have covered you before are already sharing it. Momentum is building; this is the one to push while it is live.
For standup. A national daily ran a favorable piece on your category that puts you in good light. Useful context, no action today.
Skip. The rest is routine syndication and reposts. No new outlets, no sentiment swing on anything you flagged as sensitive.
7:03, the follow-up
The pickup: how does it compare to our last launch? Same outlets, or new ones?
7:03, the answer
Broader. The last launch stayed in trade press. This one is being shared by mainstream business reporters too, several who have never covered you before. A new audience is opening up, worth a tailored pitch today while interest is fresh.
That is the whole scan. Total elapsed: under two minutes, including the time it took to type the second question.
Yes, this is a clean run, and I picked a good morning to show you. The point is not that every answer lands this tidy. It is that the triage happens at all, inside the window you already have, and that you can argue with it in plain language. It runs the same when the news is bad: on a crisis morning it points you at the fire first, with the same single question.
Why this works (and why it is not just “summarize my inbox”)
If the value here were just summarizing 180 mentions, it would be table stakes. Every monitoring vendor has summarization, the digest in the inbox is proof, and it has been there for years without changing the judgment part of the morning. Three things make this different.
Weighting, not counting. The assistant did not return the top 20 by reach. It surfaced the one item most likely to matter today, based on how fast it is moving, who is engaging, and how the last comparable story played out. That is a judgment, not a sort, and a scheduled digest cannot make it for you.
Cross-tool reasoning in one breath. When the follow-up asked how the pickup compared to the last launch, the assistant pulled the previous launch from Determ’s history, checked the CRM for which reporters the team already knows, and flagged the ones who are new. Three data sources on one desk, which is the thing the MCP does.
A recommendation, not a report. “Worth a tailored pitch today while interest is fresh” is a next step, not a chart. You can disagree with it. You can also just do it, and that is the difference between a tool that informs you and a tool that takes a task off your plate. It does not make the call. It clears the operational and administrative weight, the sorting, the cross-referencing, the first pass, so your judgment lands where it matters.
Worth saying out loud: this is the part a demo cannot show you. The demo shows the prompt. The morning shows what happens after it, when you ask the second question. I know because I run a version of this on our own coverage every morning before standup.
What it needs from you
None of this is magic, and it is not plug-and-walk-away. The triage is only as good as the setup behind it. Here is the honest list of what the workflow needs from you.
A clean topic setup. The assistant only reasons over what Determ is tracking. If a topic, market, or competitor is not set up, it will not be in the answer. Coverage quality is on you, and it is worth getting right before you lean on the morning brief.
Your definition of high stakes. Tell it which outlets, accounts, and topics carry weight, and the ranking sharpens immediately. Skip that, and it treats a regulator and a random repost as equals. A few minutes of context up front change everything downstream.
Context for the bigger questions. Ask it “what story should we tell next quarter” cold and you get something generic. Give it your messaging guide, your targets, and your last few campaigns, and the same question gets a useful answer. The limit is not the model, it is how much of your world it can see, and that is something you can fix.
A human on the call. It ranks, compares, and recommends. Deciding what to actually do, and owning it in the room, stays yours. The workflow gets you to the decision faster; it does not make it for you.
These are not reasons to wait. They are the reason teams who set it up well get a different morning from the ones who bolt it on and hope.
How to try this without changing your stack
If you already use Determ, ask us for MCP beta access in the chat on determ.com. The step-by-step setup guide walks through Claude, ChatGPT, and the rest. Setup is a one-time job: about two minutes in Claude, three in ChatGPT, and the others use the same URL. After that, the only thing that changes is which window you open in the morning.
New to Determ? Start a free trial, and once your account is running, ask us for MCP access. The short version of what it does lives on the Determ MCP page.
No Determ yet, just curious? The shape still works. Try this with whatever monitoring data you have today, even a manual paste:
“Here are the last 24 hours of brand mentions. Weight by source authority and reach concentration. Tell me the one thing worth acting on before my 9 a.m. standup, and what is safe to skip.”
You will not get the cross-tool reasoning without a real integration, but you will get a feel for the shape. And if your monitoring tool cannot talk to your AI assistant at all, that is worth a question to your vendor.
What I am doing next
I am writing a series on working with AI over your media data, the honest version. Next up: getting Claude set up as your daily PR tool, the projects and prompts that make it stick. After that, building a knowledge base that actually knows your business, so the assistant answers like someone on your team rather than a stranger. Then a full competitive media analysis, end to end, in about 30 minutes.
If there is a workflow you want me to write up next, reply on LinkedIn or send it to me.
Ivor Bihar is COO at Determ. He co-writes this series with Claude as a drafting assistant. The product decisions, opinions, and workflows are his. Company and outlet details in the scenarios are illustrative; the data shapes and conversations are drawn from real beta use. How the workflow works →
