Qase Without the Login Screen: What MCP Adoption Actually Looks Like

There's a version of this post that just says "MCP adoption is up and to the right." That's true, but it undersells what's actually happening. Historically, if you wanted to log a result, update a test case, or check coverage in Qase, you opened a browser tab and logged in. Qase was a destination. Increasingly, it isn't — it's a capability that shows up wherever the work already is: inside an IDE, inside an AI coding assistant, inside a CI pipeline. The person is still doing the work. They're just not making a special trip to do it anymore.

The short version: since launching MCP the week of March 23, 2026, the number of distinct teams calling Qase from somewhere else, instead of logging in directly, has grown from 107 to over 1,150. And it's still climbing.

The electric grid had this exact shape

Before the 1900s, if a factory wanted electric power, it needed its own generator, usually a steam-driven dynamo in the basement. Power wasn't something you called on, it was something you had to own, staff, and stand next to. Every factory was its own destination for its own electricity.

Samuel Insull changed that. Instead of every building running its own "isolated plant," Insull built large central power stations and ran distribution lines out to everyone. For factories, streetcars, homes , electricity simply arrived wherever a machine already was. Nobody had to visit the power station. You didn't even think about where the power came from. You just plugged in the sewing machine, the lathe, the streetlight, and it worked. Rates fell, adoption exploded, and the shift had nothing to do with replacing the machinist or the seamstress, they were still doing exactly what they'd always done. What changed was that the power now came to them, instead of them having to go and generate it themselves.

That's the shift our MCP data is showing. The same testing work, requested from wherever people already happen to be working, instead of requiring a separate trip to a login screen.

The numbers

Since the week of March 23, 2026:

And as of this week, the cumulative count has already passed 1,168 teams.

It's not a flash in the pan

Percentage growth naturally decays as the base gets bigger, week 2 was +111%, week 16 was +6%. That's just arithmetic. What's more telling is the absolute number of new teams adopting MCP each week, which has held remarkably steady for the last two-plus months: roughly 55–80 net-new teams, every single week, since mid-May. This is a consistent, compounding drip that adds up to hundreds of teams per quarter.

Teams aren't just trying it, they're routing more of their real work through it

This is the part a pure "number of users" chart misses. Divide total weekly call volume by the number of onboarded teams, and the average intensity per team has roughly doubled, from about 45 calls/team in week one to over 100 calls/team by week sixteen (a rough proxy, since not every onboarded team is active every single week, but the direction is unambiguous). Teams aren't just connecting it once out of curiosity. More and more of their day-to-day testing work is arriving through this channel instead of through the browser.

And it isn't a narrow slice of read-only lookups. The two most-called endpoints are for individual test case operations and test case listings, and the method breakdown shows a large and fast-growing share of PATCH calls alongside GET. Essentially, people are using our MCP to create and update test cases, not just glance at coverage. It's the same work a person would otherwise do by clicking around the Qase UI, just requested from wherever they already were: their editor, their AI assistant, their pipeline.

The tell: usage doesn't stop when people log off

Over the most recent week, daily active teams held around 300–330 on weekdays and dropped to 44–56 on Saturday and Sunday. That's a real, expected dip, because most testing work is planned during the week. But it isn't zero. Dozens of teams and thousands of calls are still flowing through on a Saturday.

That's exactly what you'd expect once a capability is wired into pipelines and tooling rather than gated behind a login. A CI run kicked off earlier in the week can still report results back on a Sunday; a scheduled job can still check coverage overnight. Nobody needs to be at their desk logging into Qase for that request to reach us. The request was already set up as part of the team's normal workflow, and it just keeps running.

Both, not either/or

None of this is an argument against the UI. If anything, it's the opposite. Our interface is consistently one of the things customers point to when they explain why they chose Qase over the alternatives: it's fast, it's clean, and plenty of teams genuinely enjoy planning runs, reviewing dashboards, and triaging results inside it. That hasn't changed, and MCP adoption isn't replacing it. In fact, the two are growing side by side.

The real value isn't "headless instead of the UI." Teams now get to choose, moment to moment, which one fits: log in when you want to sit down and work through a test review properly, or call it from your IDE, your AI assistant, or your pipeline when you just need a fast answer without breaking your flow. Same electricity, same appliance and plenty of people still like flipping the switch by hand, and that's exactly as it should be.

Why this matters

Insull's power grid didn't spread because he replaced factory workers but, rather, because the same work people were already doing got easier the moment power stopped requiring a special trip to go get it. Our MCP adoption curve looks the same. It’s not people being taken out of the loop, and not the UI being phased out, but the same testing work reaching us through whichever door people choose, whether it's a browser tab or a tool they already had open. If the last four months are any indication of the next four, that shift is still just getting started.