An Expert’s Take on How to Get Real ROI from AI

Why PMs must define their value beyond coordination

With AI, the role of a project manager either moves upstream or dies.

I don’t spend time coordinating tasks, chasing people, or babysitting trackers anymore. AI handles status updates, meeting notes, follow-ups, scheduling, formatting, and first-draft reports — faster and cleaner, without complaining. That work no longer defines value.

Here’s the warning most PMs miss: If that’s where your value lives, you’re in trouble.

Here’s the warning most PMs miss: If your value lives in doing the work, you’re in trouble.

Share This Quote on:

• Share on Twitter
• Share on LinkedIn
• Share on Facebook

Justin Bateh

PhD (Operations Management), PMP

What gets my attention now are the things that drive outcomes:

• Framing the real problem
• Naming a single owner
• Surfacing risk before it explodes
• Forcing decisions with clear options and consequences

I use AI to strip out noise and surface signal early, so I can step in sooner, align the right people, resolve tradeoffs, and set direction. My value isn’t in doing the work. It’s in making the calls that move the work forward.

Where impact lives for AI-enabled PMs

As coordination and execution get easier to automate, they matter less. But judgment still requires a human:

• Decisions
• Power dynamics
• Knowing which risks matter and which don’t
• Framing tradeoffs
• Reading incentives
• Knowing when to push, when to wait, and when to blow something up.

Automation cannot handle that slice of work. That’s where real impact lives. If you’re not operating there, you’re not indispensable. You’re just next.

Why PMs should stop running projects the "right" way

I used to run projects the “right” way. Big platforms. Long plans. Beautiful status templates. They took a lot of work and achieved very little progress. I spent most of my time feeding the system to make it look controlled. That was the mistake.

I shut it all down. I kept only what was still alive and rebuilt with a lighter, harder-edged setup. Notion holds context and priorities. Excel runs the business — decisions, owners, risks, commitments. Zapier handles handoffs so nothing stalls. ChatGPT does the grunt work like meeting summaries, action lists, and first-pass updates. No chasing.

I used NotebookLM to scan the past so I didn’t lose real decisions, but I didn’t drag old habits forward. If it didn’t matter, it didn’t survive.

The biggest challenge of this shift was surprising: It was ego. I was used to systems that looked impressive and felt safe, even when they slowed everything down. When I stripped things back, there was nowhere to hide. No bloated plans. No dashboards pretending we had control. That made people uncomfortable — fast.

The next problem was resistance. Some folks equated fewer tools with less rigor. I handled that by being ruthless about clarity. One owner. One decision. One next move. No debate.

The final challenge was letting go myself. When AI took over coordination, I had to stop hovering. I put in simple checkpoints, trusted the signal, and moved on.

The result is fewer meetings, shorter updates, lower overhead, and problems surface early, allowing for timely resolution. And no one asked for the old system back.

How AI is redefining core project delivery rituals

I define scope on one page, in plain language, tying it to a decision and an owner. If it can’t be explained without jargon, it’s not scoped. AI strips it down, but I make the call.

For alignment, I don’t run kickoff theater. I map owners, decision rights, and incentives upfront and use AI to surface conflicts before the work starts. If alignment depends on “ongoing communication,” it’s already broken.

For validation, I kill long review cycles. I set clear acceptance criteria, use AI to do first-pass checks, and bring in humans only when judgment is required.

Execution runs on simple rules. AI summarizes progress, flags stalls, and surfaces risk. I step in when a decision, escalation, or reset is needed.

The pattern is consistent. AI does the busywork. I do the thinking. And anything that doesn’t drive a decision, reduce risk, or move the work forward doesn’t survive.

Justin's Tips

Define scope on one page, in plain language, tying it to a decision and an owner. AI can strip it down, but you make the call.

A real-world example of AI simplifying project delivery

Here's an example of AI simplifying project delivery. We had a team buried in noise. Endless updates. Conflicting priorities. Risks showed up late, and everyone acted surprised.

The system was the problem. So we ripped it apart.

Our first move was consolidation. We put all project material into one place: NotebookLM. Docs, notes, emails, decisions. No debate. This took a couple of hours and immediately eliminated the “I didn’t know” excuse.

Next, we used ChatGPT to enforce discipline. Weekly updates, risk summaries, and stakeholder briefs all followed the same structure. No creative writing. No spinning. Prep time dropped from hours to minutes.

Then, we put Excel back where it belongs: running the business. Decisions, owners, risks, and commitments, with simple rules that flagged stalls and upcoming calls. Nothing fancy.

The result wasn’t magic. It was control. Risks surfaced earlier. Meetings got shorter. Leaders stopped asking for random updates. The team stopped reacting and started making decisions. This stabilized delivery and rebuilt trust.

Why tools aren't responsible for your AI-integration issues

When adopting AI in our delivery workflows, we were surprised to find that the problems weren't technical at all. The tools worked. The workflows worked. Our thinking didn’t.

AI exposed that fast. Unclear decisions resulted in garbage output. If no one truly owned something, it showed up immediately. There was nowhere to hide.

Once we clarified goals, ownership, and tradeoffs, everything else snapped into place. The real shift wasn’t speed. It was discipline. We stopped reacting late and started making clean decisions early. That changed how the work moved.

Once we clarified goals, ownership, and tradeoffs, everything else snapped into place. The real shift wasn’t speed. It was discipline. We stopped reacting late and started making clean decisions early. That changed how the work moved.

Share This Quote on:

• Share on Twitter
• Share on LinkedIn
• Share on Facebook

Justin Bateh

PhD (Operations Management), PMP

Why PMs should cut tools that create motion without progress

Over the last year, the win hasn’t been more tools. It’s been fewer. I cut anything that created motion without progress.

I’ve also walked away from legacy project management platforms entirely, including the ones slapping AI on top and calling it innovation.

Here's my AI stack:

• ChatGPT is the tool I touch first. Not to write; to think. I throw messy inputs at it, force decisions into the open, and pressure-test plans before they leave my desk.
• Claude comes out when the stakes are higher. Long documents. Sensitive language. Situations where tone matters and sloppy wording creates risk.
• NotebookLM is the project’s memory. It's the quiet weapon. I load the real documents so we stop arguing about what we think was decided and refer to what was actually decided. That alone has killed a lot of rework. The payoff is fewer alignment meetings, tougher conversations that land faster, and far fewer mistakes about who matters and who doesn’t.
• Gamma is for speed. When something needs to be visual and out the door, fast.
• Zapier keeps things moving behind the scenes so updates don’t stall.
• Notion is just the home base. No ceremony.

Why NotebookLM is a game changer for project delivery

I want to emphasize the importance of NotebookLM in project management.

We had a project that kept “needing alignment.” That was the lie.

The real problem was nobody knew who decided anything. Every meeting recycled the same opinions. Approvals dragged for weeks.

I stopped the debate. I pulled the real artifacts: Emails, decks, notes, old approvals. I loaded everything into NotebookLM. Then I asked blunt questions:

• What decisions were made?
• Who was in the room?
• Who approved versus who just talked?
• Where did things stall?

The output was uncomfortable. Two so-called decision makers had never approved a thing. One person with real power had quietly fallen out of the loop.

We rebuilt the stakeholder map around reality, not titles. We cut two approval steps. And we put decisions back in the hands of the person who owned the outcome.

The result? Meetings disappeared. Decisions landed in days instead of weeks. The project moved because we stopped guessing and started dealing with facts.

When project leaders should avoid using agentic workflows

I'm starting to experiment with agentic workflows, but I'm being deliberate about it. In fact, I just signed up for a course on agentic workflows and orchestration platforms to go deeper before broadly rolling anything out.

Right now, I’m focused on the safest parts of delivery to automate first: coordination, handoffs, summaries, and signal detection. Tasks that create drag but don’t require judgment.

I’m not trying to automate decisions or leadership. That’s a mistake.

So far, I've found value in understanding where agents can reduce friction without adding complexity. It's early, but the direction is clear. Use agents to compress noise and surface signal, not to replace accountability.

Why AI will split project management in two

I think project management will split in two — and there’s no middle ground. One side will be automated. The other will be elevated. Most people won’t like where they end up.

This rule quietly governs project delivery in the AI era. The closer my work is to decisions, the harder I am to replace. The closer my work is to tasks, the easier it is to automate. This has nothing to do with effort, credentials, or experience. It’s about proximity.

Most project managers today focus on task coordination. I did for years. Coordinating work. Chasing updates. Tracking timelines. Making sure nothing falls through the cracks. Critical work. And almost all of it can now be automated. AI already owns execution. The coordination game is over.

The worst response I see is doubling down on execution. Better trackers. Cleaner reports. Tighter process. That’s not strategy. That’s denial. You’re competing where AI wins every time.

Execution work implements what’s already been decided. Decision work shapes what gets decided in the first place. One happens downstream. The other shapes the stream. Only one holds durable value.

Execution work implements what’s already been decided. Decision work shapes what gets decided in the first place. One happens downstream. The other shapes the stream. Only one holds durable value.

Decision-shaping means identifying decisions no one wants to make. Connecting issues across workstreams. Translating technical problems into business consequences. Surfacing risks leadership hasn't seen yet. Framing tradeoffs before execution locks them in as a strategic thinking partner.

This is the line that determines everything for me. If my value shows up after decisions are made, I’m downstream and exposed. If I shape what gets decided, I’m upstream and durable.

AI is compressing value toward decision work whether people like it or not.

Why delivery leaders must embrace AI-driven change

Here's my advice:

• Stop protecting the work that AI automates. If most of your pride comes from being organized, responsive, or “on top of everything,” you’re defending the wrong hill. AI already does that work better and cheaper. Let it go.
• Move yourself upstream on purpose. Start showing up where decisions are framed, not just executed. Don’t wait to be invited. Bring leaders the decision they’re avoiding, the risk they’re missing, or the tradeoff they don’t want to name.
• Kill anything that doesn’t force clarity. If a meeting, report, or ritual doesn’t end in a decision, an owner, or a consequence, it’s theater. Cut it.
• Use AI aggressively, but not blindly. Let it handle coordination, summaries, and signal detection. Save your time for judgment, influence, and accountability. That’s the work that compounds.
• Finally, be honest with yourself. If your value shows up after decisions are made, you’re exposed. If you help shape what gets decided, you’re durable. Choose accordingly.

Follow along

You can subscribe to Justin Bateh's free weekly newsletter, Tactical Memo — weekly tactical briefs for project managers and operators working in the AI era. Each issue tackles one execution problem and delivers one clear solution.

More expert interviews to come on The Digital Project Manager!