Your Inbox Is About to Become Someone Else's Problem

Had a random thought today about email.

With Copilot, Claude, Gemini, and others getting deeper into our inboxes and actually acting like EAs (if you set them up right), it seems like we're heading toward a world where we spend more time talking to our assistant about email than actually being in our inbox.

Not quite there yet. Mine still misses stuff I'd consider important, and even a small miss rate feels too high given what's at stake in some threads. But I do think that gap closes in the next 2–3 years.

Agents Talking to Agents

It's easy to imagine a shift where a lot of "emailing" isn't really human-to-human anymore. More like agents handling the back and forth, translating intent, sending, receiving, organizing. Leaving more time for us to think strategically about the highest-value work.

We're already seeing companies like Firecrawl, LangChain, and others focused on making the internet easier for LLMs to read and understand. It's only a matter of time before something similar happens to email itself — not just reading your inbox, but actually understanding the relationships, priorities, and context behind every thread.

But Here's the Part Nobody's Talking About

If AI agents start handling the coordination layer — the scheduling, the follow-ups, the "just checking in on this," the status updates — then what happens to all the systems we built specifically because humans are bad at coordination?

Think about Jira. Think about Monday.com. Think about every ticketing system, every project management tool, every standup meeting. These exist because humans need to serialize and hand off work to each other. We need a shared artifact — a ticket, a card, a thread — to make sure nothing falls through the cracks.

But what if agents are handling the back and forth natively? What if intent gets translated directly into action without needing to be written down in a format that seven other people can parse?

The ticket becomes overhead. The standup becomes a formality. The project board becomes a relic of a time when coordination required human labor.

Who This Actually Impacts

This isn't theoretical. If you think through who spends most of their day translating intent between systems and people, the list is long:

Project managers. Not the strategic ones who are making hard calls about scope and priority — they'll be fine. I'm talking about the ones whose primary function is making sure person A knows what person B said, and that the Jira board reflects reality. If agents can do that natively, the role changes fundamentally.

Executive assistants. The best EAs are already thinking three steps ahead of their exec. But the ones who are primarily managing calendars, drafting responses, and sorting inboxes? That's exactly what these AI agents are being built to do. The role doesn't disappear — it evolves. But it evolves fast.

Customer support. Tier 1 support is already getting automated. But when agents can actually read, understand, and respond to customer emails with real context — not canned responses, but actual understanding of the customer's history and sentiment — that changes the entire support org structure.

Sales ops and account management. So much of sales follow-up is mechanical. "Checking in." "Following up on our last conversation." "Wanted to share this resource." If an agent can handle the cadence and you just handle the relationship and the close, that's a different job.

Middle management broadly. Any role where the primary value is information routing — taking input from one group, reformatting it, and passing it to another — is going to feel this. Not tomorrow, but soon.

The Uncomfortable Part

I want to be clear: I'm not saying these roles go away. I don't think that's how this plays out. But I do think the shape of these roles changes dramatically. The people who thrive will be the ones who were already doing more than coordination — the ones making judgment calls, building relationships, seeing around corners.

The coordination layer was never the valuable part. It was the tax we paid because humans can't share context the way machines can. Once that tax goes away, the value has to come from somewhere else.

What I'm Watching

I don't think this is five years away. I think we're 2–3 years from AI agents that can genuinely manage your inbox, your calendar, your follow-ups, and your project status — well enough that you trust them with it. Not perfectly. But well enough.

And when that happens, every tool and every role built around human coordination overhead is going to have to answer a hard question: if the coordination is free, what are you actually selling?

I'll probably have more to say on this as I keep thinking about it. For now, it's just a thread I'm pulling on.