Project Management in the AI era

Every time a new tool comes out, someone writes that the Project Manager role is going to disappear. They said it with Jira. They said it with Notion. They're saying it now with Copilot, Claude and the agents. And they're wrong every time, but a little less wrong with each cycle. Because AI doesn't eliminate the role, but it does mercilessly shut out PMs whose value was compiling status reports and dragging tickets between columns.

What is changing, and fast, is the shape of the work and the line between what adds value and what is noise. Here's what I'm seeing in real projects.

What AI is absorbing from the classic PM role

There's an entire layer of administrative Project Management tasks being automated and, frankly, it was about time.

Summaries and meeting minutes

Hand-written minutes have stopped making sense. A Teams recording, a decent transcript and a well-built prompt produce minutes that are more complete, more useful and faster than most PMs can deliver in 30 minutes. The PM's value isn't in writing the minutes anymore — it's in deciding which decisions deserve a follow-up and which don't.

Weekly status reports

If your status board is still being filled in by hand every Friday, you're losing 2 to 4 hours a week that AI can deliver in 10 minutes from Jira, Asana or Monday. What the senior PM adds is the reading: what risks are hidden in data the report doesn't capture, which stakeholder needs a call before Monday, and what number is being polished without anyone saying so.

Plan and backlog drafts

To kick off a project, AI can spit out a reasonable first plan of tasks, dependencies and rough estimates in less than an hour if you give it a good brief. It used to take me a whole day. Now I spend that day questioning the plan, not producing it.

The first plan is no longer the deliverable: it's the starting point. The PM's value has shifted from production to critique.

What becomes more important (not less)

As AI absorbs the administrative layer, what stays visible and critical is the part that should always have been the heart of the role.

Decision under uncertainty

A serious project isn't a sequence of tasks, it's a series of hard decisions under incomplete information. AI doesn't do this for you, and won't for a long time. When to cut scope, when to escalate to committee, when to say no to a sponsor, when to accept technical debt: that's the raw material of the senior PM.

Political reading

In any mid-sized or large organization, projects are won or lost in conversations that never appear in the official plan. Who's uncomfortable, who feels skipped over, what direction a silent sponsor is leaning. AI can help you remember who said what, but not read the room. That fully human skill is worth more than ever.

Tailored communication

AI writes correct emails. The senior PM picks when email isn't the right tool and a five-minute call works better, or a hallway conversation, or a well-placed silence. Knowing what channel and what tone fits each counterpart is what separates someone who leads from someone who only coordinates.

The new PM profile that's taking off

Over the last twelve months, I've seen an emerging profile I call "augmented PM". They have three traits that weren't required before and now are.

AI fluency

Not "knowing how to write prompts". Fluency of use: keeping the tool open like you keep your calendar, knowing how to ask for comparative analyses, risks, first drafts, translations. The augmented PM saves between 6 and 10 hours a week with this, and invests them in decision and conversation.

Data and metrics as a native language

The augmented PM isn't satisfied with "we're behind". They look at real velocity, lead time, throughput, blocker rate. And, critically, they know when a metric is measuring real work and when it's theater. This, which was bread and butter in my time at Block, is now going mainstream.

The ability to simplify

Since there's infinite information available, the PM who can prune and simplify is the one who survives. Saying "this doesn't fit in this sprint" or "this report doesn't deserve to exist" requires earned authority and judgment. Those who do it well rise. Those who add noise to look busy fall.

The minimum kit for a PM in 2026

I'm not talking licenses, I'm talking habits.

• A go-to conversational AI open all day (ChatGPT, Claude or Copilot — whatever the company has).
• A note-taking tool with automatic meeting summaries (Granola, Otter, or whatever's integrated).
• Properly configured access to Jira/Asana/Monday and to a basic BI to read reality without waiting for the data team to run a query for you.
• A private Slack or Teams channel where you keep useful prompts, templates, lessons learned.
• Twenty minutes a week of "sharpening the axe": trying a new technique, reading a short guide, watching how others do it.

Real risks of the PM who delegates too much to AI

Not everything is positive. I've seen cases where AI made the PM's work worse, not better.

The automaton PM

Some PMs have delegated so much to AI that they no longer think — they just forward what AI produces. The team notices fast and stops respecting the PM, who turns into a facade. If you save hours but lose authority, the trade is awful.

Loss of instinct

When AI summarizes everything, you stop reading the fine print. And projects usually die in the fine print. It's worth maintaining the discipline of reading two or three key documents in full per project, even if AI gave you the summary.

Overconfidence in estimates

AI produces estimates with a confident voice. That doesn't make them true. I've seen plans presented to committees with AI-generated numbers that nobody validated. When the plan falls apart, the PM can't defend an estimate they didn't believe themselves.

The mistake I see most often

The mistake I see most often is the PM who uses AI as a quantity amplifier: producing more documents, more summaries, more updates, more plans, more everything. They end up drowning the team and stakeholders in information no one asked for. AI shouldn't turn you into a page factory; it should turn you into someone who delivers less, better.

The rule I apply: every time AI generates a deliverable for me, before sending it I ask whether the recipient actually needs it or whether I'm sending it to look productive. Half the time, I delete it or cut it in half. That discipline, in a world of infinite output, is what separates the senior PM from the noisy one.

The PM in the AI era isn't less important. They're more exposed. AI comfortably covers 30-40% of the administrative tasks that used to be a refuge. What's left is what was always the heart of the craft: deciding, reading, communicating, sustaining. If your value was there, this amplifies you. If your value was compiling reports, it's worth rethinking your direction sooner rather than later.