If you only watched the keynotes and read Microsoft blog posts, you could conclude that Dynamics 365 is now primarily an AI vehicle, framed around Copilot features and agent scenarios designed to revolutionize sales and service work. That’s quite natural. In marketing, it’s always about what’s new. Oh, and in certification exams, too.

The functionality we as customers and partners of MS bizapps work with on a daily basis doesn’t change with every release wave. We rely on the core of what systems like CRM were built to handle. We build on top of what the apps have, by either adding or removing pieces to make it fit the purpose of the system. That foundation doesn’t change radically — not even in ten years.

But a lot will happen within a decade. Evolution made us humans pay attention to the sudden movements, which was a crucial skill to keep us alive in the jungle. We’re not as well equipped to notice the slow changes that gradually change everything around us. The creeping climate catastrophe being one concrete example of this.

Recently, on a couple of occasions I needed to compare what Dynamics 365 is today vs. what it was a decade ago. This made me realize how much I take things for granted now that were not available in 2016. We should remember to celebrate the progress that we’ve seen in things that truly did matter. Likewise, it helps in calibrating our minds about the importance of shiny new things when we acknowledge that some of them will not matter much a few years later.

That’s what this week’s Perspectives issue is all about. Let’s begin our 2016→2026 journey!

2016 was one of those big “reimagine” moments at Microsoft. Dynamics CRM was replaced with Dynamics 365. Microsoft’s CRM and ERP were to become one business suite to rule the world. “Microsoft Dynamics 365: One Cloud” was one of the marketing slides.

Just because it didn’t necessarily go like the slides promised, doesn’t mean it wasn’t significant. The Dynamics products were taken to the mainstream of Microsoft’s offering, alongside the booming Office 365, which clearly resulted in more investments into R&D. The pace picked up considerably from there on.

Reimagining Dynamics 365 as a set of apps you could pick and mix rather than a single thing called CRM was controversial for existing customers, sure. But the upside from it all was that this made a real business application platform possible. Apps became something everyone could build. Whereas XRM certainly wasn’t an environment you’d just let anyone have a go at.

In the classic Dynamics CRM world, you would customize the system. Even in the modern Power Platform Environment Settings app, you can still see the legacy terminology in there. It remains only there for backward compatibility purposes, offering a route to the CRM 2011 era Solution Explorer.

The term “customize” assumed that you bought a system for your business and then you wanted to change it or extend it. In the Power Platform era, we no longer do that. We can of course customize existing apps from Microsoft, or RapidStart, or whoever is the solution vendor. But if we choose to build from scratch, the tooling is exactly the same. It is today a platform for both makers and vendors. This is the single biggest conceptual difference to how things were in 2016.

If we ignore the crazy naming adventures, the foundation formerly known as XRM is still the most critical piece of Dynamics 365 CE/CRM. For a history lesson, check out my article from last year.

Sure, it’s still a database — among other things. What started as the single DB per CRM server (until multi-tenancy in CRM 4.0) evolved first into organizations and then Power Platform environments. If you’ve ever seen me preach talk about low-code governance, you’ll know that environment strategy is at the core of what makes it real. The services and tooling that Microsoft has built on this front over the recent years has huge value for all customers, regardless of what data sources their apps and automations use.

The idea of Dynamics being basically just forms over data was somewhat valid ten years ago. Today, you’re missing out on most of the platform capabilities if you stick to that mental model. Even though the big ambitions around virtual entities tables never materialized, the power of connectors changed the game from moving data around to reading/writing data against various sources on the fly.

The connector capability comes from the Power Apps/Automate side of the house. When the Power Platform became a thing in 2018, a big reason for merging XRM with PowerApps & Microsoft Flow (as they were called) was to avoid reinventing the enterprise wheel. The simple apps and flows needed a lot of tooling around them to make things scale. Luckily, much of that was already up and running in the Dynamics 365 cloud.

The needs for creating a platform for one billion apps from all makers rather than just a few predictable CRM systems pushed the offering forward. One side was the self-service nature of how Dataverse environments can be used today. Yes, not all of the simplified modern UIs are very power user friendly. Luckily, behind the UI there’s been plenty of progress.

Enterprise Dynamics CRM projects had attempted to follow common software development practices already before Power Platform. Yet most deployments were never at the scale where investing in “proper” SDLC methodology was financially justifiable. As a result, consultants and admins just customized systems each in their own way, through the available means and with the resources available to them.

Starting from a business application platform means you’re not really doing custom software development the way the teams at Uber or Spotify are. In 2016, most teams doing Dynamics 365/CRM work didn’t resemble a classic dev team at all, even if some members had aspirations to adopt practices from the big boys. The concepts, the tools and the business needs were just too far apart for much of those to be applicable in everyday consultant life.

ALM was always a concern, though. Just because you didn’t write software from scratch didn’t mean that getting the app updates reliably and repeatably deployed into the hands of business users was easy. CRM folks knew the risks and issues, they just lacked the tools that would have been fit for purpose. As the solution framework launched in 2011 became the model offered also for individual citizen developers, this only made the divide between theory and practice wider:

In the above article I wrote six years ago on this topic, I longed for a built-in deployment pipeline like the one Power BI had introduced. Fast forward to today and Power Platform Pipelines are a reality. And in general, they work! That doesn’t mean Pipelines are perfect for everyone, but I mean: OMG, things are so much better for low-code ALM now! We even have native Git source control integration in Dataverse.

The pros don’t probably like the restrictions of built-in Pipelines and that’s just fine. Deep down, the features are part of the same APIs that PAC CLI calls. Which means you don’t need a GUI, you can do things with code. And in 2026, anything that’s expressed in code is something LLMs are well equipped to tackle. See where this is going? That’s right: the terminal as the new UI (gift link to Plus article is here).

For power users and low-coders who aren’t afraid to open VS Code and talk with GitHub Copilot, the world of platform APIs is now open. The way how Power Platform Connectors democratized talking to other SaaS apps (even if it’s essentially “credentials sharing as a service”) is now happening to any endpoint.

For as long as there’s a human in the loop, the visual side of the UI still matters. If we look at what the first-party Dynamics apps for processes like Sales and Customer Service are today, the main ingredients are still the same as ten years ago. Today we just call them model-driven Power Apps in technical terms. The detailed recipe has evolved substantially, though.

Splitting CRM into semi-independent app product teams meant not every app needed to look the same. Domain-specific UI controls like the pipeline view in Dynamics 365 Sales Hub became an everyday thing as the product experience was allowed to diverge.

“Aren’t we now trapped inside non-standard UI experiences that can’t be tailored for real customer needs?” On some level that is true, but it’s not that bad actually. Thanks to it all being built on top of Power Apps Component Framework, everybody can now build these non-standard controls in a supported way. A decade ago, the idea of replacing any standard control with your custom one was something the PMs at Microsoft quietly mentioned to us at the MVP Summit presentations in Redmond. Today, PCFs are an everyday technology in the platform.

In 2026, the popular hot take on social media is “canvas apps are DEAD!” In 2016, we didn’t have them on the same platform to begin with. What is dead for sure is embedded canvas apps on model-driven forms, given how Microsoft has since then prioritized custom pages as the technology for pixel-perfect UIs inside model-driven app frames. Used extensively in both commercial products as well as Kits from Microsoft, it’s hard to see these going away soon — no matter how slick React apps you can create with vibe coding today.

Code Apps recently went GA and provide an interesting option for building completely custom experiences on top of Dataverse + the connectors via prompting your AI tool of choice. Whether that would ever turn into productized Dynamics 365 apps that customers are expected to “vibe customize” remains to be seen. Nevertheless, in a world where UIs can become entirely generative, the value of a stable data layer and identity + access management tooling may prove to be significant.

The one area where it’s difficult to find any major leap forward is ironically what business apps like Dynamics 365 should really be about: business process automation. Storing data is one part, showing it on nice form and grids is another one - but what does the system actually do with that data? Or how can it help users do the thing they need to do, with less… “doing”?

The core capabilities of XRM plug-ins and workflows have not been revolutionized by the cloud. For cross-system process automation, yes, we’ve definitely gained plenty of new tools to connect apps and data sources. But for in-app logic, there isn’t much progress on the platform level that could unlock more complex process automation, calculations or process UX support.

The first-party Dynamics 365 apps have each gone and built automation features specific to their domain. It’s sort of natural that the needs of a customer service rep are different from those of a sales manager or a field engineer — resulting in diverging app experiences. The IPR of Dynamics products has been moved into containers protected by the app licensing, to more clearly distinguish them from the custom Power Apps that you could be running at $20/user/month. When paying $105 for Customer Service Enterprise, for example, you obviously need to get some bang for the buck beyond just basic ticket queues.

When it comes to platform-level functionality like Business Process Flows or Business Rules that were already available ten years ago, Microsoft lost their ambition to push things further. They never developed tools for more complex BPM scenarios, even though the technology would be right there in the house, like the MS Visio integration of AgileXRM demonstrates. The slow advancement of Power Fx support or the back & forth with low-code plugins is a sign of there being a lack of direction for deterministic enterprise automation within Microsoft. All the internal incentives are encouraging taking the non-deterministic Copilot route.

At one point it seemed like Microsoft was launching new Dynamics 365 products every month. This was driven by the combination of A) having a fairly mature cloud business application platform, B) existing customer base for the core CRM products, and C) the excitement around tech trends like Internet of Things, Mixed Reality and Artificial Intelligence. A + B + C = potential new revenue to go after and make BizApps a bigger slice of the MS pie. Considering Microsoft was never the number one in neither CRM nor ERP for enterprise software market, it made sense to try new things that built on top of their uniquely broad tech stack.

In 2020, the website for Dynamics 365 showed a dizzying array of applications available. The number of Insights products kept expanding and you could have organized a bingo to see who guesses the next app to be launched: “Customer Insights!” “Customer Service Insights!” “Product Insights!” “Sales Insights!” “Market Insights!” “Finance Insights!” “Insight Insights!”

Some got to a point of commercial availability, resulting in Microsoft’s local sales offices to invite customers and partner to immerse themselves in the Digital Feedback Loop powered by Dynamics 365 cloud, Azure IoT, Hololens, all that exciting, shiny tech. Others were scrapped before leaving preview. In any case, practically nothing from that booming era remains as a meaningful product used by customers today.

Marketing is perhaps the exception here. It’s an area where Microsoft was struggling massively ten years ago already. They had to go through plenty of iterations, renaming, merging and reimagining, yet at least today Customer Insights is a thing actual organizations are using. Journeys (the marketing part) bundled together with Data (the Customer Data Platform) is an offering that addresses a real need. It’s hardly the market leader in this game, nor suitable for any SMB customers of Dynamics 365, yet it has its place.

All the other new products may have been a false start for the AI and data era. You could argue, though, that it’s a necessary step in the evolution of business software. People say many of the product ideas that crashed and burned in the dot-com boom of 2000 have since then evolved into viable digital business models. The first AI boom of Dynamics must have cost Microsoft some money, sure — yet it was an entirely different ballpark than the runaway capex spend we see in today’s AI bubble. If lessons were learned from the “Insights era” of Dynamics, perhaps it was all worth it?

In the year 2026, we’re still searching for a meaningful way to turn AI into a product Microsoft can sell and customers can benefit from. Unlike a decade ago, customer decision makers are actively asking for AI tools that could help their sales, service and marketing employees achieve the kind of magical results that tech media and LinkedIn influencers keep shouting about.

You’d imagine that at this point, after all the experimentation and investment, there would be a product that one could confidently offer to these customers as “here, take this, it’s a solid choice for most your needs”. But no. Whatever it is today, it will be a different thing tomorrow.

When I started with Microsoft Dynamics CRM 3.0 exactly two decades ago, the CRM Outlook client was a major selling point of the software. No, it wasn’t smooth sailing from a technical perspective, but at least the technology didn’t change every year. In 2017, Microsoft first announced they were going to deprecate the thick client and push everyone to the Dynamics 365 App for Outlook. Which is what they eventually did get to doing in 2020. But just look at what has happened with the proposed successor of the App for Outlook since 2022:

In addition to these “products”, we of course now have the agents as a new dimension. A recent document from Microsoft explaining the different sales-related use cases of agents included in total 15 agents — just for Sales!

That’s a list provided by Claude. I literally now have to ask an LLM parse the information coming from Microsoft. Ten years ago, I could have just bought Dynamics CRM 2016 and start installing it on my Windows server. I was a big advocate for moving to the public cloud because things were much more likely to work there without technical issues blocking the end users and system customizers. Today, the blocking issues are more about not understanding what’s A) actually available in the cloud, B) knowing what it will cost, and C) predicting whether it will be there tomorrow still.

In times like these, it gives some comfort to know that there’s a robust business application platform behind it all. That you’re diligent about how to manage customer data in a compliant way. Instead of, you know, just uploading a list of PII as CSV or Excel files into an AI chatbot session and hoping for the best… Whoops!!!

In March 2026, Microsoft’s first and so far only A365 agent is still lacking an integration into a CRM system. It seems like this wasn’t as simple to implement as the product marketing team thought, but the pressure to release Frontier agents meant it had to be launched. There’s also a testimonial from Microsoft themselves how during a pilot program in Feb-Jun 2025, Sales Development Agent “engaged more than 70,000 existing Microsoft SMB customers”.

To put things into perspective, it was probably fine to conduct this agentic AI pilot with real contacts in the US. Had Microsoft attempted to do something like that in the EU, the word “fine” might have switched from an adjective to a noun. At least if the quick comparison done with the help of my AI assistant on this CRM scenario is accurate:

To end this on a positive note: Dynamics 365 remains a solid foundation for addressing crucial business requirements around CRM processes. It may not ship with all the necessary bits out of the box, yet Power Platform today has extensive tools for making the applications work the way modern business technology should. When you can both protect the core of what a CRM system must deliver day-to-day, as well as support new experiments that are necessary for companies to remain competitive in their markets, you’re in a good place.

It’s okay to get excited about all the vibes going around, and how easily we can generate entirely new apps and agents. Ultimately, I don’t think that’s what LoB systems like Dynamics 365 are about. It’s also the reason why they don’t necessarily get better at what they do by just adding more of new AI on top.