By Pierre Hulin

Every so often, project management changes shape. A scheduling technique from the late 1950s grew into software that runs entire portfolios. Mainframes gave way to client-server, then to the cloud. Task lists became strategy maps. Each time, the people running the work had to rethink what their tools were for.

We're at another of those moments. Artificial intelligence (AI) is changing what a portfolio system can do on its own, and the question leaders ask is moving from "what can I find out?" to "what should I already know?" Before looking ahead, it's worth looking back at how PPM got here, and what autonomous portfolio intelligence means for the organizations running portfolios today.

By autonomous portfolio intelligence, I mean a system that continuously monitors the portfolio, surfaces what matters to the people who need it and suggests what to do next. The monitoring is autonomous. The decisions are not: people still make the call.

Where it all began: a short history of project portfolio management

Project portfolio management (PPM) is older than most people think. The scheduling methods at its heart date back to the late 1950s. DuPont and Remington Rand Univac developed the Critical Path Method between 1956 and 1959 (PMI), while the US Navy developed PERT to manage its Polaris missile program (PMI).

Over the following decades, resource management, cost management and portfolio views were layered on top of the schedule. But scheduling never went away. It was the original pain point PPM was built to solve, and it's still central to PPM and SPM products today.

"At the core of it, PPM really was scheduling."

The market has reinvented itself several times along the way. PPM moved from mainframes to client-server in the 1990s, IT PMOs became early adopters, and cloud-native vendors arrived from the early 2000s. Collaborative work management (CWM) grew up alongside it, and in 2020 Gartner described the PPM market splitting into strategic portfolio management and adaptive project management (Gartner).

Strategic portfolio management (SPM) is now a category in its own right, with Gartner's Magic Quadrant for Strategic Portfolio Management most recently published in June 2026.

That history matters, because it looks nothing like the evolution of AI. PPM is a mature market that goes back decades. AI, as most people now think of it, means large language models, a space that only really took off when OpenAI released ChatGPT in November 2022. The interesting question is what happens when a decades-old discipline meets a technology that's less than four years old.

Same core problem, new goalposts

Strip away decades of added features and PPM still answers the same questions. What's in the portfolio? Who's working on what? When will the milestones land, and what will it cost?

What has changed is the role of strategy. It was always part of the picture, but it was rarely formalized. Today, more organizations run a dedicated strategy office, sometimes called a strategy realization or value realization office. Linking strategy to the portfolio is now part of what organizations expect from PPM.

Why even good strategic portfolio management is still a reactive loop

SPM promised to connect delivery to strategy, and at its best it delivers. You can see your goals, see the work that supports them, and make informed calls: drop projects that don't serve the strategy, or reshape the portfolio when priorities shift toward a new business unit or market.

But look at how that works in practice. The strategy changes. The change trickles down to the portfolio. Someone makes the adjustments, then goes back up to check that everything still lines up. Every step starts with a person asking a question and the system answering it.

That's the limit of even the best SPM. It's a much better way to answer questions, but someone still has to ask them, at the right time, about the right projects.

Autonomous portfolio intelligence breaks that loop.

"You set the goalpost once, you ask the question once, and then the system continuously monitors the portfolio to ensure you're tracking along the direction of travel."

The question becomes a standing instruction. The system keeps checking the portfolio against it and tells you when something starts to drift.

What is autonomous portfolio intelligence?

Most enterprise PPM platforms are systems of record. They hold structured data and run business workflows on top of it. The catch is that they're very rigid and very reactive, not proactive. People have to go looking for information. Even the most advanced platforms mostly raise a flag through an alert or a report, then wait for someone to act.

Autonomous portfolio intelligence turns that around. The system continuously monitors the health and status of the whole portfolio, surfaces what matters to the people who need it, and suggests what to do next. People still make the call.

I see this as a natural next step. As organizations mature, they want to move from static, rigid systems toward something dynamic that gives every stakeholder more proactive options.

Where autonomous portfolio intelligence matters most

Continuous monitoring matters most where portfolios are complex and tightly controlled. In aerospace and defense, capital projects and life sciences, programs run for years, depend on one another and operate under strict governance. A schedule slip or cost overrun that goes unnoticed for a full reporting cycle can ripple across the whole portfolio. A system that keeps watch, and raises the issue as soon as it appears, gives teams time to act while they still have options.

Taking portfolio data from the PMO to the boardroom

PPM has long been seen as an operational tool that lives inside the project management office (PMO). I believe autonomous portfolio intelligence moves it up to the boardroom.

The reason is simple: executives and board members don't log in to systems. Their attention is pulled in every direction. They expect a briefing prepared for them and delivered to their inbox, or whatever channel they use, weekly for executives, monthly or quarterly for board members.

Building that briefing has traditionally taken a lot of human effort, with analysts mining data, finding the signals and polishing the output. Intelligent systems can increasingly do that work and deliver the signals straight to key stakeholders.

The benefit goes beyond the board. Subject-matter experts who were pulled in only when something went wrong can stay in the loop all the time, with far less friction. I expect that to improve data quality, spread knowledge across the business, reduce risk and speed up delivery.

Board reporting in hours, not weeks

Executives have always wanted information yesterday. The difference now is that it's becoming possible.

Early in my career, I worked with customers who spent a month to six weeks pulling together information for their annual budgeting and board meetings. A well-designed PPM platform cut that to a few weeks, sometimes days. With autonomous portfolio intelligence, it comes down to minutes or hours, mostly spent reviewing the data.

People stay firmly in the loop. Today, human review must cover how an AI agent reached its result, as well as the result itself. Hallucinations are real, and the industry is at the very beginning of this journey.

As organizations run enough evaluations to trust their agents, I expect that review to change shape. People will check the output without having to retrace every step behind it. And the decisions that matter, such as approving, canceling or funding projects, stay with people.

Bolted on or built in - why AI-native PPM matters

Almost every vendor now attaches AI to its product. So why isn't autonomous portfolio intelligence just "AI-powered PPM"?

I'd point back to the shift to cloud. Cloud-native vendors disrupted the market and grew fast. Legacy vendors spent years reworking their architecture to host software in the cloud, and the result was often less fluid than a product built for the cloud from day one.

I see the same pattern playing out with AI. Most vendors that claim to be AI-powered are just legacy vendors that have bolted AI onto their platform. That's a step in the right direction, but it isn't real innovation.

Think of companies that grow by acquisition. Different architectures and data models create friction that's expensive to remove, and users feel it in convoluted workflows and workarounds.

The signal to look for is an AI-native product, where AI can touch every part of the system. That's what makes always-on AI agents possible: agents that scan the data and proactively alert people and suggest actions, instead of waiting for someone to come and look. The human is still the one who decides.

AI in portfolio management: from production to proven value

We're seeing more and more people shifting from proof of concept (POC) to production. Many customers are now in production with AI, past the experimentation and minimum viable product (MVP) stage, often with multiple AI systems and processes in place.

The next step is value. It's one thing to say that you have systems in place and you're beyond the POC. Being at the stage where you can measure the value that it has derived is kind of the next step. Measuring return on investment (ROI), and even measuring the cost, is still at an early stage, and that's where the market is heading next.

The next chapter for project portfolio management

PPM began as a way to keep a few critical projects on schedule. Nearly seven decades later, scheduling is still at its core, and the question it answers hasn't changed: what are we working on, and is it taking us where we need to go?

What changes now is who does the asking. Our view is that the next generation of portfolio management will be judged on three things. Does it watch the portfolio continuously, instead of waiting for someone to ask? Does it keep people in charge of the decisions that matter? And can it show, in hard numbers, the value it delivers? Most of the market isn't there yet, and that's what makes this moment worth paying attention to.

Sources