Process Mining is now Process Intelligence

Last week Gartner published the 2026 Magic Quadrant for Process Intelligence platforms. Until this year it was the Magic Quadrant for Process Mining platforms. The vendors and their positions are roughly where you'd expect. The more interesting change is the name on the report itself.

The origins of process mining

The term “process mining” was coined in 1999 by Wil van der Aalst and Ton Weijters at Eindhoven University of Technology, in a research proposal called “Process Design by Discovery”. The idea was a small one with big implications. Stop separating how a process is designed from how it actually runs. Use the data left behind by real executions to learn what the process really looks like.

That academic origin matters, because process mining is genuinely a specific analytical technique with formal foundations in computer science. It's not a category of dashboards or a flavour of business intelligence.

Process mining was always a means to process intelligence

Of course the technique was never the point. The point was always to understand how your business runs and make more intelligent decisions because of it. But there was a stretch of years where it was useful to put the name of the technique on the box. When buyers saw process maps and conformance metrics for the first time, they needed to know why this was different from their existing business intelligence platforms. “Process mining” was the answer. It told them that something specific was happening under the hood, something with academic rigour behind it which made it different from any other analytical investment they'd already made.

For buyers at the top end of the market, this explainer is no longer needed. The leading vendors have been selling process mining for well over a decade and process mining is now established as an analytics category in its own right. Putting “intelligence” on the box reflects what was always being purchased.

The choice we made

The term “process intelligence” has been floating around for some time. It's the obvious word for the obvious thing, in the same way business intelligence was the obvious word for what data mining produced twenty years ago. With the generative AI surge came the slogan “no AI without PI” which was popularised by Celonis and is now being used everywhere. It seems to have played a role in solidifying process intelligence as the category's new label.

My co-founder and I spent a long time discussing which term to adopt for our public content. Process intelligence was the trendier term, the one the market was moving towards, and there was a clear bandwagon to jump on.

Would “process intelligence” risk us sounding like yet another startup chasing the latest AI trend? Which is awkward, because AI is central to how Flow Myna works. It's what makes process mining outputs accessible to an operations manager who has a factory to run.

We chose process mining. Our target audience, SMB manufacturers, are not the FTSE 100 buyers who already understand what is under the hood. They still need an explainer, not a slogan. “Process mining” tells them that something specific and analytical is happening to their data.

Our vision for Flow Myna hasn't changed. But Gartner's rename has reopened the question of what words we use to introduce ourselves to a buyer who's never heard about any of this.

What is new is what process mining output gets used for

Process mining has been feeding downstream systems for years. Human analysts first, then automation tools consuming discovered process models to drive RPA and process orchestration. The newest systems are LLM-powered, and that's what “no AI without PI” is really getting at.

Now of course there's plenty of AI without process intelligence. An LLM helping you code or draft an email to your manager doesn't need to know how your processes run. But if you want copilots answering questions about your operations, agents completing tasks and orchestration engines making decisions, then the underlying LLMs need context grounded in the specifics of how your operations actually run.

Process mining produces exactly this context, in a form that is both machine-readable and human-readable at the same time. A process path discovered from your data is, almost uniquely among analytics outputs, a story: an ordered sequence of naturally described events. And that's the kind of structure LLMs are fluent in.

Process mining was always producing intelligent output. But there has been a real step change in who and what can consume it.

Final thought

The starting point is the same. You take event data, you reconstruct what's actually happening, and you produce process intelligence a business can act on. That hasn't changed.

What has changed is accessibility and reach. Five years ago the intelligence ended in a dashboard for a process mining analyst. Today it can feed automation tools, copilots, and agents. Tomorrow there will be yet more places to go. But it all still starts with process mining.