App Dev for MSPs: Insights On Using Copilot

Microsoft partners are beginning to realize that new opportunities they once dreamt of are now being created!
When cloud came along and pretty much crushed hardware sales and the attached services, many partners started to try to figure out what they should do next. One of the obvious possibilities was to get into app development.

But let's face it, most said," Nahhh", just as soon as the idea occurred to them. There was just no way they were going to be able to develop coding skills fast enough to build a business around it. Some who may have thought that they could hire those capabilities instead said not likely. The world still has millions of job requests open for developers, but there are a precious few candidates out there. 

For myself, I remember often explaining to customers that we could provide the canvas, but we didn’t do the painting. “We could recommend someone!” So, it seemed, AppDev was just not in our future.

A recent study in TechXplore states that, “AI is already writing almost one-third of new software code ” with other reports citing higher percentages and pretty much everyone agreeing that this trajectory is heading sharply upward.

One of the foremost AI experts, OpenAI co-founder and former Tesla senior scientist Andrej Karpathy, has recently said, “I rapidly went from about 80% manual + autocomplete coding and 20% agents to 80% agent coding and 20% edits + touchups. I really am mostly programming in English now.”

When even the foremost coders are no longer coding, the possibility of Microsoft Partners expanding their abilities to include application development totally changes. Some partners realized this during the low-code/no-code era that they, too, could create applications with these tools. With AI-enabled coding assistants, their ability to create applications for customers suddenly becomes much more available.

When was the last time you entered commands into a command line interface (CLI)? What was the last batch file you wrote? When have you obtained and configured an API key? Client IDs? Application IDs?

The most valuable answer to all these questions is “the next time I do any of these, I will learn or re-learn how to do them,” and you’ll be right. Most experts agree that the easiest way to learn anything about working with AI is to work with AI. Learn by doing.

The story Microsoft is telling about Copilot and developer productivity is more nuanced and more useful for MSPs than the headline “55% faster developers.” It’s really about changing where developer time goes, how fast new people ramp, and how teams feel about their work. 

If you talk to customers about Copilot, the first number they’re likely to have heard is that GitHub Copilot makes developers 55% faster. That comes from a tightly controlled experiment: developers were asked to build a small HTTP server in JavaScript. Half had Copilot, half did not. The Copilot group finished in about 71 minutes; the control group needed 161.

That’s a great proof point, but it’s not real life. In the real world, developers are juggling meetings, reviews, investigations, production issues and a constantly shifting backlog. Microsoft’s more recent work, including diary studies, telemetry, and the Copilot Usage Report 2025, tries to understand what actually changes once Copilot is in daily use.

What they found is that the “speed” story fragments into a few different dimensions that MSPs can actually sell: time saving for repetitive work, higher perceived productivity, better flow and faster ramp‑up for new hires.

So where does Copilot fit in? Copilot shows up first and most often in the boring parts of the job. In both Microsoft and third‑party studies, developers report that AI assistants save time on boilerplate, repetitive patterns, tests and glue code. When Copilot is on, they spend less time hunting Stack Overflow, less time typing obvious code and more time on the parts of the problem that are actually specific to the business.

Second, their experience of work changes. Microsoft and GitHub consistently see developers saying they feel more productive, more satisfied and more able to stay “in the flow.” In surveys, large majorities say Copilot helps them get started faster, reduces mental fatigue and keeps them in the IDE instead of bouncing between browser tabs. That doesn’t immediately show up as more pull requests per day, but it absolutely shows up in whether people feel like they’re getting somewhere.

Third, onboarding curves bend. Microsoft’s own internal research suggests it takes around 11 weeks of daily use before the hard productivity metrics begin to move. In a three‑week randomized trial of Copilot use inside Microsoft, telemetry (time coding, PR counts) didn’t change much, but participants reported that Copilot was already helping them work faster, especially on boilerplate and unfamiliar APIs. The implication for MSPs is simple: if you deploy Copilot and walk away after 30 days, you’ll miss most of the value.

For MSPs, the practical takeaway that “AI improves developer productivity” is not a one‑line promise; it’s a services opportunity. Someone has to do the hard work of turning Copilot from a shiny license into a new way of working. Microsoft’s usage data reinforces this: Copilot shows up in about half of developers’ workdays, and mostly on code‑heavy tasks where it’s been intentionally woven into the workflow.

That creates a consulting lane almost tailor‑made for you. You can help clients identify the high‑leverage use cases, boilerplate, tests, API exploration and documentation, then redesign the process around them, pairing Copilot with code review practices, validation and secure coding standards. You can also design rollout programs that last long enough for teams to get through that 8 to 12 week learning curve, instead of stopping at the procurement phase.

There’s also a governance angle that enterprises are only starting to grasp. Microsoft’s own three‑week study highlights a key tension: AI lets you generate code faster, but that code still has to be validated. Developers reported that some of the time savings were eaten by additional review effort, especially when Copilot produced code that “looked right” but actually wasn’t. When an MSP brings opinionated patterns here, such as what gets auto‑generated, what is always human‑reviewed and where to use tests as the arbiter, they're not just selling seats, they’re lowering risk.

The bottom line for your customers is this: Copilot is already shifting how Microsoft’s own engineers and millions of developers work, but the gains are uneven and depend heavily on how teams adopt it. Your role as an evolving MSP is to turn that messy reality into a disciplined practice your clients can rely on, with clear use cases, measurable time savings on the right tasks and a culture where AI is a force multiplier rather than a source of subtle bugs and false confidence.

So yeah. AppDev is definitely within reach for MSPs.

Posted by Howard M. Cohen on March 30, 2026