"What's in My Inbox?" — The Question You'll Want AI to Answer

Happy Wednesday! This week:

Let’s get started!

Communicators have a peculiar problem: there's almost never a shortage of work.

Most comms teams are thinly staffed and pulled in five directions at once. One minute you're on an executive announcement; the next you're weighing in on a product issue, answering a reporter, rewriting a deck, or building talking points for a meeting you didn't know was on the calendar. And you're rarely brought in at the start — you get looped in after the decisions are made, with the context scattered across someone's inbox and the key call buried in a meeting transcript. "Can you just take a quick look at this?" and congratulations, you've acquired another project.

Sound familiar? We end up tracking an enormous number of moving pieces, half of which aren't technically ours — until they are.

And that's before the work nobody ever explicitly assigned you. We’re also tracking down execs who haven’t gotten back to us yet, following up with reporters or internal stakeholders, and trying to remember to run things by the brand and legal teams, and a lot, lot more. And most of this happens via email. Which is both the problem — and, with AI, an avenue toward a solution. Why? Because your inbox isn't a communications channel anymore. It's a task-management system you never designed.

Well, not never designed — haven’t designed yet.

Our inboxes are effectively infinite and never close — anyone can ping us about anything at any time. No human was built to hold that many open loops in their head at once.

This is exactly the kind of problem AI is good at, and almost nobody is pointing it there. We keep asking AI to write things. The bigger unlock is asking it to tell us what we're supposed to be writing in the first place. I know this one works because I live in it — I built myself an AI chief of staff to do exactly this, and it's changed how I work more than any drafting trick ever did. Here’s a practical outline to get you started.

If you're mostly using AI to summarize and draft, start here.

Feed it your recent inputs — a batch of emails, your meeting notes, a Slack export — and ask it to find the loose ends. Questions you haven't answered. Requests you haven't fulfilled. Commitments you made, and commitments other people made to you. Decisions still waiting on a follow-up. A deadline you may have blown past. People sitting there waiting on you. Something that's quietly escalating.

The shift is small but it's the whole ballgame: don't ask AI to summarize — ask it to detect what's missing. A summary tells you what a message says. You already read it. You want to know what it obligates.

If you're already comfortable letting AI analyze your information, move from detection to prioritization.

Take that pile of loose ends and have AI turn it into an action list — what it is, what to do, how urgent, why, and the very next step. Here’s a sample output; you can give the column headings to your chatbot so it gives you structure back.

The point isn't to hand AI the wheel, but to make it answer the question every communicator is silently asking at 4 p.m.: of all the things I could be doing right now, what actually needs me?

If you're already analyzing and prioritizing, don't stop at a to-do list. For each item, ask AI to produce the first usable piece of work. Not the last 20%, but the first twenty.

The goal here is not "AI does my job." It's this: AI erases the blank page between "I need to do this" and "I've started doing this." That gap — the flinch before a task you haven't opened yet — is where half your loose ends go to die. Close it and the whole day moves faster.

This is the rung most people never reach, and it's the one I find most valuable. Once AI is helping you handle individual tasks, ask it to manage the system.

At the end of each day or week, hand it everything that came through and ask: What's still unresolved? What am I waiting on from other people? What are other people waiting on from me? What did I commit to that has no evidence of being done? What should I carry into next week?

Now AI isn't processing information for you. It's maintaining continuity — the thread of what's open and what's owed that usually lives in one overworked person's memory and walks out the door the week they're on vacation. For a communications team, that's worth more than another press release. It's the difference between using AI as a faster typewriter and using it to run a workflow.

But wait — don’t do the above yet. You need to build two habits before you go down this path.

1) Check with cybersecurity.

“Dump your whole inbox into a chatbot” is terrible advice. And I want to be clear that I am absolutely not saying that — full stop.

Confidential information, privileged communications, and personal data don't belong in a tool your company hasn't cleared — ever. AI is no different, except that it takes the accidental disclosure problem and makes it a billion times worse. You need to make sure you have permission to share with your chatbot before you share with your chatbot.

The good news: in many cases, this is a lot easier than it sounds. If your organization runs the Microsoft suite, there's a good chance Copilot is already wired into your email and calendar and can do much of this with no copy/pasting at all — the inputs are already right there. Elsewhere, you may be able to set up a connector that lets Claude or ChatGPT read your inbox directly. Either way — and I can't say this loudly enough — clear it with your IT or cybersecurity team before you point any AI at real company information, and use the tools your organization has approved.

2) Verify everything that AI hands back.

Everything. Dates, deadlines, who-said-what. And keep the judgment where it belongs. The most important line here: AI should surface possible obligations, not invent them. If it tells you a reporter is waiting on a reply, go confirm the reporter is actually waiting. The tool raises its hand and says "this looks like it needs you." Deciding whether it really does is still the part only you can do.

If you’re handing off AI-assisted work without poring over it, you’re engaged in malpractice. I know this should go without saying, but I italicized “should” there because it often doesn’t.

Don't take my word for any of this — test it. Take the emails you received over the last week, the ones you haven't fully dealt with, run them through an approved AI tool, and ask three things:

If you're in a Copilot shop, you may not even have to paste anything — just ask. Then notice how long it took. I'd bet it's faster than the hour you were dreading, and I'd bet it surfaces at least one thing you'd forgotten.

Because here's where I've landed. We spend all our energy asking AI, "Can you write this for me?" — and it turns out that's not even the interesting question. The one that actually moves the needle for communicators is quieter, and a little uncomfortable: What am I supposed to be doing that I haven't done yet?

Join Fin and Plaid on August 13th to see how AI can resolve issues directly in customer conversations. This includes everything from reducing bank-linking friction, enabling proactive support, and helping customers complete real financial tasks.

Reserve your seat

Here's the uncomfortable part of the AI-slop conversation we haven't had yet: you can produce nothing but careful, human-made work and still get punished for it. All it takes is running an ad next to someone else's synthetic garbage.

That's the finding buried in DoubleVerify's new 2026 Global Insights: Media Quality in the Age of AI, a survey of 22,000 consumers across 22 markets and 2,020 marketers — big enough that the numbers are hard to wave off. Globally, 42% of consumers say low-quality AI advertising would sour their opinion of a brand, and more than half of marketers are already worried about their ads landing next to low-quality AI content. That’s right — it doesn’t have to be your AI slop that hurts you. Being next to someone else’s could do similar damage.

But — and this is the part that should reframe how we talk about it internally — this isn't an anti-AI story. The dividing line isn't whether you used AI. It's whether what shipped was any good. A bit more than half of consumers say AI tools improve their online experience, and 46% feel positively about ads inside AI platforms. They’re cool with AI — they're rejecting low-effort use, a/k/a slop. That's a far more defensible position for a comms leader to hold than a blanket "authenticity" argument that treats AI as inherently cheapening.

So what do we actually do about it?

Whether you're trying to build systematic AI adoption across your team or focused on AI search visibility for your brand or organization, I’m here to help.

I have three slots open for late August/September — learn more and book an intro call.

Most AI resources are scattered, untested, or buried behind a paywall.

The Shift puts it all in one free newsletter, covering what’s new, and what practically works under 5 minutes a day.

A vault of 3,000+ vetted tools so you’re never guessing. A 1000+ prompt library for every use case you’ll actually hit and free AI courses that make sure nothing stays theoretical.

Plus, right now, 3 subscribers win a free 1-year Claude Pro subscription just for signing up.

Subscribe to The Shift

Was this forwarded to you by a friend or colleague? Want to get this in your inbox next week? Click here to subscribe!

Quick question: how can I help?

What workflows are you struggling with? Where does AI still feel mysterious or overwhelming? What has worked that you’d like to share with others?

I’m a builder, and I’d love to help you and the rest of The Comms Stack community find great new ways to use AI.

Reply and tell me.
I read every response.

Even a one-sentence reply helps. For example:
“I wish AI could help me with ______.”

Until next Wednesday,

Dan

Advertise | Subscribe