Some Microsoft products are able to stand the test of time and avoid the perils of getting “reimagined”. Cornerstone software like Excel or PowerPoint has remained mostly intact for decades now. Meanwhile, other products that might at first seem like cornerstones of information work have not been so lucky. They get tossed into a limbo where nothing is certain - what the products are called, what features they support in any given incarnation, whether there is a roadmap or not.

I wrote about Microsoft Planner and the revenge of MS Project a few months ago. This article touched upon the strange evolution of what is today called Planner Premium and the customer confusion caused by decisions of MS. Once customers of Planner Basic discover that Planner Premium isn’t just offering more features but rather pushing them into a completely different platform, not everyone (anyone?) is happy about it. As a result, customers are looking to downgrade from Planner Premium plans and restore core functionality.

Project and now Planner have a relationship with Power Platform that has made me observe the product evolution, albeit from a distance. Business apps like Dynamics 365 Project Service Automation (PSA) and later D365 Project Service have been products I perhaps should have dived deeper into. Yet anytime I’ve touched them, I’ve been glad that I’ve avoided them. I still have fond memories of dodging a bullet back in 2017 when a customer and our sales team asked me to deliver a PSA deployment and I said, “we don’t have the skills for this”, in my brutally honest and often controversial way for a MS consultant.

I’ve since then continued to explore the strange fate of Microsoft Project in a broader context. The many different audiences that MS is trying to cater for, with no consistent masterplan, is an intriguing story. As we now see even the core Office apps experience being sacrificed on the altar of Copilot, the lessons from Project can help us in putting these seemingly irrational decisions into broader context.

Historically, Microsoft Project was undeniably powerful. Its robust scheduling engine, resource planning capabilities, and fine-grained control made it the logical choice for corporate PMOs. But while Excel and PowerPoint thrived through incremental improvements as the modern work expectations evolved, Project stagnated.

That doesn’t mean we wouldn’t have seen new releases in this field. On the contrary, Microsoft introduced various cloud flavors for PM: Project Server, Project Online, and eventually Project for the Web, each one promising innovation for whatever was the hot tech thing at the time. The natural end result for customers was confusion on what to use where, thanks to the combination of overlap and feature gaps.

The latest move of rebranding Project for the Web as Planner Premium only compounded customer frustration. Casual Planner users found themselves forced into complex Dataverse-based environments, losing straightforward integrations they had grown accustomed to, like embedding in SharePoint or simple Power Automate flows with readymade connectors. Professional project managers, meanwhile, felt betrayed by the simplistic features of Planner Premium compared to Project Online or the classic desktop MS Project.

MSP veterans have shared their frustration online. Dan Swaigen, a Project consultant with 21 years of experience with the software, recently summed it up on LinkedIn in his “R.I.P. MSP” post. Dan described Microsoft Project’s current state as a "dysfunctional family" of tools, plagued by unclear naming, fractured licensing, and a lack of meaningful updates. His analogy of “Saying you can use MSP to do agile is like hitching 1000 horses together to compete in an F1 event” captures the awkwardness of Microsoft’s attempt to retrofit legacy tools to modern methodologies.

Microsoft has announced that Project for the web will be retired in August 2025. From there on, users visiting project.microsoft.com will be redirected to Planner. Considering what the experience of accessing an iconic domain like that is today, the situation could hardly get worse. Because right now you may be directed to at least four different user experiences of project management apps when clicking on a project on that page:

Does removing one option from the list solve this chaos? If anything, the 2019 product of P4W is the easiest for Microsoft to eliminate from their commercial portfolio. The underlying gaps between Planner Basic and Planner Premium (aka P4W) will not be eliminated via rebranding, so MS still have their hands full with trying to make the Planner story and feature set logical for customers.

Being a low-code advocate and an early Power Platform evangelist, it is sad to see the brave initiative of replatforming the earlier SharePoint based Project products onto Dataverse and Power Apps stalling. I guess the lesson here is that no technical platform choice can save you from the impact of misaligned product management strategy. The newer tech just introduced new challenges instead.

Back when P4W launched, customers of Project Online were urged to migrate to it. Today, as P4W gets put to rest, the classic Project Online experience will remain in place. Because as is often the case in Microsoft’s product stack, the modern thing can’t replace the classic thing. Thanks to their global presence inside businesses who are paying for licenses, Microsoft has a tough time pulling the plug on classic products without offering a migration path. They’re not quite like Google, thank heavens.

So, why did Microsoft give up on Project? To understand decisions of MS today, we can utilize Occam’s razor: the simplest explanation being most likely the right one. When it comes to Microsoft in the year 2025, the explanation we should fall back on is that anything favoring the rise of AI is the decision MS teams will select.

The Project Manager Agent in Planner was first demonstrated at Ignite 2024:

After waiting for a few months, I finally saw the features light up in my tenant. It helped in understanding why MS had chosen to stop investing in the professional PM tooling and instead was chasing the lightweight task management and casual project coordination needs. It’s simply because they want their software not to manage the work but also do the work.

In Microsoft’s vision, GenAI agents will do “research” using content from MS Graph and deliver Loop files that humans are expected to review and collaborate on. When leveraging the Project Management Agent for a project like “build and deploy a website”, the agent jumped from the planning phase into execution phase almost by accident. There are big prompts for “assign to Project Manager” in the UI that encourage the plan owner to throw things at AI and see what comes out from the other end of the pipeline.

In theory, it’s the dream of what businesses want computers to do. The whole work, rather than just helping human employees manage the work. An automated labor factory that analyzes the requirements given, plans the tasks to be done, and then does them.

Does it work? Well, if vibe-coding is a hot topic, surely vibe-managing a project is an equally attractive concept for people to explore. Yet just like there aren’t any serious (let alone secure) vibe-coded software products out there, it’s not very likely we’ll see any critical work in projects being fully completed by LLM based agents in the near future.

On one hand, a lot of everyday project management is about dealing with the soft data that comes from people. Messy, incomplete, conflicting, fragmented, misplaced, delayed, duplicated… I do see the potential for taking a large language model and giving it a task to process this information. For getting close enough summaries for non-critical tasks and unfamiliar topics. That’s exactly where I personally leverage GenAI every day.

But I don’t trust it - because I know it’s not trustworthy. The bigger the models get, the better they get at lying to us. It’s not a bug, it’s a feature of LLMs. It’s not likely to get solved by doing more of the same training that gets us ever growing reasoning models. This is why I’ve created the following test for detecting suitable scenarios for applying GenAI tools:

You can easily use AI to help visualize this concept and check whether you would react differently to an app that was more honest about what it is underneath the shiny tech speak:

This gets us to the issue with Microsoft Project and how I see its foundation to be at odds with what the GenAI layer is doing. The arcane project scheduling service that needs to be used still today for getting accurate task scheduling results for Planner Premium tasks has been impossible for Microsoft to “reimagine” in a cloud native way. Because they need it to be accurate. The only way to not break the existing scheduling logic has therefore been for MS to essentially run a winproj.exe instance in the cloud. Think about that for a moment.

Obviously, something as complex and critical as this is no place for Random Dude As A Service. This is the other side of project management where accurate calculations are needed. It’s something computers used to be good at - until we invented a way to make them creative in a non-deterministic way via LLMs. If only we could clearly separate these two different patterns of computing from one another, things might work out. But both the tech vendors and also the end users are not equipped to preserve this separation.

I regularly see very smart and knowledgeable people share their discoveries about how Copilot has been able to complete some tasks they gave it. They get excited about how AI responded to them in a convincing, human way, and want to share it with the community. Then, I look at the data and numbers in the response and see that they are wrong. AI didn’t succeed in the task - it only succeeded in pretending that it did. This always makes me think about what a massive challenge it is for all of us to protect ourselves from the persuasion and manipulation that the LLM based services excel in.

Can something that’s called “Project Manager” be inaccurate and unreliable? This is the fundamental dilemma I see with the PM agent in Microsoft Planner. Are we safe from getting drawn into the illusion where we believe that the AI agent is able to do everything it claims to be? Including accurate processing of data and calculations? No. I believe the typical human project manager is not prepared to deal with these hallucinations at all.

Humans are not perfect - but they can be held accountable. In fact, I’d argue that the primary reason for a project manager to exist is that someone needs to be accountable for the project’s progress and outcomes. Traditional software like Microsoft Project is not able to do much on behalf of the human user, yet it is very reliable in what it does. One could say that the classic PM software and its vendors have been accountable for their part. As they now try to step beyond that, there’s a vast ocean of conflicts and distrust ahead of us.

The frequent problem with Microsoft’s products is that the Redmond organization can only ever focus on one truth at a time. Yesterday it may have been collaborative apps in Teams, today it’s Copilot agents. The official narrative is tailored to serve the needs of rolling out the latest shiny object, with no references or even acknowledgement that the reality of business users is not as black/white as the product marketing story.

It feels to me like Microsoft have decided that since Project never evolved into a suitable tool for agile project methodology, they’re going to skip right past agile and go to what’s beyond that. Instead of waiting around to ask customers what capabilities they’d like to see in the apps they pay money for - Microsoft decided that they must “change the game” and build a product that serves their vision and their goals.

That is something an upstart company with a bold new vision could totally do. Disrupting the market with a new way of looking at things and challenging the status quo is how the underdogs of software business often justify their existence. Now, the problem with Microsoft is that they are everything but an underdog. As a result, they end up disrupting their own project management products - and often the workflow of their customers.

When looking at Project, Planner, and everything in between, I can’t help comparing the moves of Microsoft to someone who’s technically in the same market but doing the exact opposite: Basecamp. Thanks to how MS needs to chase every growth opportunity (like GenAI) to justify their stock market valuation to the investors, they can’t just focus on making a tool that does well what it originally set out to do. Meanwhile, a small yet profitable company like Basecamp has no pressure to sacrifice its users over higher growth metrics.

In the world of Big Tech, the dangerous moment is when products become mature enough. When there are no new products to sell to the customers. When you should improve the tools you have already sold, solving the tricky problems instead of launching the next shiny thing. That is when the interests from the customer base are in conflict with the reward system of typical software corporations that focuses on quarterly KPIs rather than long term customer retention.

This is why successful products and brands, like Project or Office, can be deliberately set on fire. Their customer base stops being the reason why the software exists. Instead, it becomes the target audience to be converted into using something new that has a higher growth potential.