/
.webp)
The number was €30 million. That was the budget. Eighteen months. Twelve teams. Seven systems. One investment bank in London that needed its trading platforms to talk to its ledger systems.
Ross Mason sat in that project and watched it happen. Three separate integration approaches running in parallel, none of them compatible with each other. File-based transfers over here. Enterprise application integration for SAP over there. Web services in the corner. Twelve teams of engineers doing what amounted to the same job three different ways, and still none of the systems could actually speak to each other in real time.
He called the work "awful."
Not the people. Not the architecture. The architecture was sound. The concepts made sense on whiteboards. But the actual software, the tools those twelve teams had to work with every morning, was inadequate for what the ideas required. The gap between what integration should have been and what it actually felt like to build it was the width of a career.
Mason would close that gap. It would take him fifteen years, cost him his mental health, and end with Salesforce writing a check for $6.5 billion.
Ross Mason was born in London. At nine, his family moved to Wales, where his parents ran a hotel. Then hotel chains. He grew up watching his mother and father operate a business where every system had to work, where a broken reservation meant a family standing in the rain at midnight with nowhere to sleep.
Two lessons absorbed by osmosis, never formally taught: never lower your bar, and working hard is not inherently negative if you enjoy what you are doing.
His grandfather was a radio ham. A tinkerer. The kind of man who built the family television from scratch and then lectured on computer theory in the 1950s, decades before anyone had a personal computer. "He reminds me that you must always keep an open mind, try things and make things," Mason said later.
For years, Ross wanted to be a Lego designer. At seven years old, he was already running a bootleg Lego club. He would have stayed on that path, possibly forever, but a Commodore 16 arrived in the house and rewired his entire trajectory. The machine was weak, slow, and beautiful. It could be made to do things. And unlike Lego, the things it did could be made to do other things.
He studied Computer Science at the University of the West of England in Bristol. Graduated in 1997.
His first real job was at an insurance company. The work was connecting over a hundred sites via dial-up connections. Unreliable lines. Dropped packets. Systems that worked on Tuesdays and failed on Wednesdays for reasons nobody could reproduce. It was miserable, tedious, and foundational. It taught him what reliability actually meant when the infrastructure underneath you couldn't be trusted.
Around 1999, Mason moved to Switzerland and entered banking. Credit Suisse. UBS. NatWest. RaboBank. Heavy back-end systems. Legacy trading platforms built in languages that predated the developers maintaining them. Data flowing, or more accurately not flowing, between systems that had been bolted together across decades of mergers and acquisitions.
Four or five years of this. He became a lead architect at RaboBank, where he built one of the first large-scale ESB implementations. Enterprise Service Bus. The idea was right. The idea was always right. Connect everything through a central bus, normalize the messaging, let systems talk without knowing about each other. On a whiteboard, it was elegant.
In practice, the tools available to build it were punishing.
Then came the project that broke him. A London investment bank. Seven systems needed to connect. The budget was €30 million. The timeline was eighteen months, which actually became closer to two years. Twelve separate teams were hired.
Three disconnected integration approaches existed in the same building: file-based transfers, enterprise application integration, and web services. None of them could talk to each other. Each approach had its own team, its own tooling, its own vocabulary. They were solving the same fundamental problem three different ways in three different rooms.
Mason had two realizations that wouldn't leave him alone. First, the integration environment was "woefully complicated," and he couldn't fully understand why it had to be. Second, the architectural concepts underneath were sound. The ideas were good. The software built to execute those ideas was not.
Something needed to exist that didn't exist yet.
In 2003, Mason started building. His first approach was app server-based. He took the integration patterns he understood and tried to implement them on top of existing application server technology.
It ran like a dog. His words. Cumbersome. Slow. It had all the right concepts packed into a vehicle that couldn't carry them. The architecture was there, but the performance characteristics were wrong. Integration needs to be fast, lightweight, and capable of handling thousands of message types without choking. App servers in 2003 were none of those things.
He nearly stopped.
Then he found Matt Welsh's master's thesis on Staged Event-Driven Architecture, known as SEDA. It was designed for high-scale web services, a way to decompose complex event processing into stages that could be independently tuned and scaled. Welsh had built it for the web. Mason read it and saw something different.
This wasn't just for web services. This architecture could apply to any distributed system. Integration was a distributed system problem. If you designed the integration broker as a staged event-driven system, it would be decoupled and scale linearly across commodity hardware. You could accept any type of event, process it through well-defined stages, and layer application-stitching concepts on top without sacrificing performance.
That was the breakthrough. Not a product launch. Not a customer win. A master's thesis from a computer science department that most people in enterprise software had never read.
He rebuilt everything around SEDA. In April 2003, he published the Mule project on SourceForge.
The name was deliberate. A mule does donkey work. Integration was the donkey work of enterprise software. Nobody wanted to do it. Nobody got promoted for it. Nobody built a career around making System A talk to System B. It was thankless, repetitive, and essential. The mule carries the load that nobody else will touch.
The open-source project grew. Developers who had been writing the same integration code over and over found Mule and recognized it immediately. Not as a product. As relief.
In 2006, Mason and co-founder Dave Rosenberg incorporated as MuleSource. The bet was the same bet every open-source-to-commercial company was making in that era: distribute free software, build a community, then build a company on top of that community with enterprise features, support, and tooling. It was the model Red Hat had proven. The model MySQL had followed. The model that worked until it didn't, and MuleSource was about to find out which side of that line it stood on.
They raised a Series A in October 2006 from Hummer Winblad Venture Partners. Series B came in May 2007 from Lightspeed Venture Partners.
In 2009, MuleSource became MuleSoft. The name was cleaner. The product was evolving. What had started as an open-source ESB was becoming something larger. The Anypoint Platform emerged as the integration layer grew into a full API management and design environment. The thesis expanded: it wasn't just about connecting systems. It was about building reusable APIs that made those connections accessible to anyone in the organization, not just the three teams of integration specialists locked in a room.
The community reached 150,000 developers. Thirty-five percent of the Fortune Global 500 were running Mule.
Total funding crossed $259 million over approximately eight rounds.
But growth is a specific kind of violence. Mason experienced depression during the scaling phase. He described it as draining. The distance between the technical founder who wrote the first SEDA implementation and the CEO of a company with hundreds of employees is not a straight line. It bends through places that nobody talks about at founder conferences.
The guiding philosophy held: "Our guiding principle is to pave roads. If something is difficult, make it easier for the next person. These paved roads take us to bigger and better places."
The roads were being paved. But the person paving them was running out of pavement.
March 17, 2017. The New York Stock Exchange. Ticker symbol: MULE.
They priced at $17 per share, above the expected range of $14 to $16. The bankers had been conservative. The market was not.
MuleSoft opened and climbed. It hit $25.92 during the day. It closed at $24.75. A 46% gain on day one. The company was valued at approximately $3.1 billion.
For context: MuleSoft had $297 million in revenue, a 73% gross margin, and was growing at 58% year over year. The numbers were clean.
The IPO raised $221 million. It beat Snap's first-day performance. An integration company. A company built on the donkey work of enterprise software. A company that made systems talk to each other. It beat the company that made disappearing photos.
Ross Mason rang the bell at the NYSE. The kid from Wales who wanted to design Legos, who got derailed by a Commodore 16, who spent five years in Swiss banks watching €30 million get spent to connect seven systems. He stood on that floor and rang the bell.
Exactly one year later. March 20, 2018. Salesforce announced the acquisition.
$6.5 billion. Enterprise value. Per share: $36 in cash plus 0.0711 shares of Salesforce stock, working out to approximately $44.89 per share. It was Salesforce's largest acquisition to date.
The logic was transparent. Salesforce owned the CRM. But CRM data locked inside Salesforce was worth less than CRM data flowing between every system in an enterprise. MuleSoft was the plumbing that made Salesforce's data useful outside Salesforce. And Marc Benioff understood something that the market had been slow to see: the integration layer was too important to leave in someone else's hands.
The acquisition closed on May 2, 2018. MuleSoft had approximately 1,200 employees. Salesforce maintained MuleSoft's independent executive, product, field, and marketing teams. It was the first Salesforce acquisition where they preserved that level of autonomy.
Ross Mason appreciated being public but welcomed the acquisition. "Salesforce is leading the charge and taking our process to the Cloud," he said. What he did not say, and what the timeline suggests, was that the IPO-to-acquisition pipeline was a twelve-month sprint that had already been running for fifteen years.
The Gartner Magic Quadrant for iPaaS named MuleSoft a Leader for nine consecutive years, from 2014 through 2024. Nine years. In the API Management quadrant, they held the Leader position for ten. For nearly a decade, if you were an enterprise evaluating integration platforms, MuleSoft was the answer Gartner gave you.
In 2025, MuleSoft dropped from Leader to Challenger in the iPaaS quadrant. Gartner cited slower innovation. Competitors like Workato and Boomi had moved faster. The architectural playbook that had defined MuleSoft, the three-tier API approach with Experience, Process, and System layers, the rigid reuse methodology, had become overhead instead of advantage.
Salesforce was undergoing its own restructuring. Five thousand roles cut in 2025. Another thousand in early 2026. MuleSoft was being folded into Salesforce's Agentforce AI strategy, positioned as the integration layer for AI agents. The independent identity that Salesforce had preserved in 2018 was dissolving.
Ross Mason wrote a farewell post on LinkedIn. Then he left.
He founded DIG Ventures. It started as a family office and grew into a $100 million fund focused on early-stage European B2B SaaS and AI companies. He became a Venture Partner at Lightspeed, the same firm that had led MuleSoft's Series B in 2007. He joined the board of Stackin' and Syncari.
He had said it years earlier, and it still held: "Entrepreneurship is a calling, an internal force that pushes one to create something out of nothing and impose that thing on the world. It requires an obsessive need to solve a problem. The amount of setbacks, silence, barriers, naysayers and the occasional kick in the groin means the only way that idea will live and grow is through an almost lunatic passion for solving that problem."
The lunatic passion built MuleSoft. A €30 million banking project convinced a Welsh kid that integration was broken. A master's thesis gave him the architecture to fix it. An open-source community proved the world agreed. A decade of grinding proved the business could work. An IPO proved the market believed. And $6.5 billion proved that the donkey work of enterprise software, the work nobody wanted to do, was worth more than almost anyone had imagined.
The first version "ran like a dog." Mason's initial attempt was built on app servers and was so slow and cumbersome he nearly abandoned the entire project. It was Matt Welsh's master's thesis on Staged Event-Driven Architecture (SEDA), designed for web services, that Mason repurposed for integration and rebuilt everything around. Most people credit Mule's architecture to Mason's banking experience. The real credit belongs to an academic paper from a completely different domain.
Ross Mason ran a bootleg Lego club at age seven. Before integration, before banking, before Switzerland, a kid in Wales charged other kids to play with his Lego collection. The entrepreneurial instinct predated the Commodore 16 that made him a computer scientist.
Mason experienced depression during the IPO-era growth phase. He described the scaling period as "draining." The gap between technical founder and CEO is rarely discussed honestly in founder narratives. Mason has been one of the few to name it directly.
The "Mule" name is a donkey joke. Integration was the donkey work of enterprise software. Nobody wanted to do it. The mule carries loads nobody else will. Mason named his life's work after the least glamorous animal in the barn.
Salesforce kept MuleSoft more independent than any other acquisition. MuleSoft retained its own executive, product, field, and marketing teams post-acquisition. It was the first time Salesforce had ever preserved that level of organizational autonomy. By 2025, that independence was gone.
Sources:
- Sramana Mitra, "Unicorn in the Making: Ross Mason, Founder of MuleSoft" (2014)
- Silicon UK, "IT Life: Taking On Development Donkey Work"
- Swisspreneur Podcast, Episode 153: Ross Mason
- TechMonitor, "MuleSoft Founder Ross Mason on Frankenstein Data Sets and APIs"
- TechCrunch, "MuleSoft soars 46% on first day of trading" (March 2017)
- CNBC, "MuleSoft IPO beats Snap in first-day trading" (March 2017)
- Salesforce Press Release, "Salesforce Signs Definitive Agreement to Acquire MuleSoft" (March 2018)
- Salesforce Press Release, "Salesforce Completes Acquisition of MuleSoft" (May 2018)
- Tom Tunguz, MuleSoft financial analysis
- HuffPost, "Going Against the Flow: Ross Mason, Founder of MuleSoft" (2017)
- Lightspeed Venture Partners, "The Past 10 Years with MuleSoft" (2017)
- Ross Mason, Entrepreneur.com columns
- DIG Ventures, TechFundingNews coverage ($100M fund close)
- Gartner Magic Quadrant reports (iPaaS 2014-2025, API Management 2015-2025)
