Chris Moustakas and the trouble with a clever idea

Chris Moustakas once explained machine learning with a jar of sweets. His ten-year-old son, Atticus, had won a school jellybean contest against more than 200 children. His guess: 1,750. The actual count: 1,729. The prize was the jar itself. Moustakas had rather hoped for a year of free tuition.

The boy had spotted a scale on the container, read the number, and allowed a little extra for the beans at the top. An ordinary measurement had supplied an unexpectedly useful clue. In his father's telling, the lesson concerned relationships in data: valuable connections can be hiding in plain sight, waiting for someone to notice them. The confectionery was a pleasant bonus.

That small domestic episode makes a useful entrance to Moustakas's working life. He has spent it building and running enterprise software, where information is plentiful, getting an answer can be laborious, and a clever tool earns its keep by helping somebody do an actual job.

A sky full of possibilities, a job full of details

Moustakas grew up in Megara, Greece, and lived there until he was ten. Outside town, the night sky was part of everyday life. Both his brothers became astronomers. At Caltech, he began with an interest in physics and finished with an astronomy degree; his public profile lists his undergraduate years as 1993 to 1997.

He joined DevonWay as an early software engineer, among its first ten employees. The company's nuclear-power work appealed to him. By 2015 he was executive vice president and chief operating officer; in 2016 he became president and CEO. The career moved from writing software to taking responsibility for the organization that delivered it.

There is a particular kind of practical intelligence in the problems he chose. A large organization has existing processes, equipment, obligations, and people who understand their work. Software has to enter that world and prove useful. A demonstration can look splendid while the daily routine remains stubbornly awkward. Moustakas's writing repeatedly returns to that distance between an attractive idea and a workable implementation.

The spreadsheet has acquired ambitions

In 2016, he wrote about what happens when powerful personal computing meets a large organization. Employees can make their own tools. They can also create disconnected stores of information, overlapping processes, and applications whose purpose becomes obscure when their owners leave. Local ingenuity can leave the next person with quite a puzzle.

His argument was that software helps people manage complexity. Giving everyone more tools does not, by itself, make an organization easier to understand. Someone still has to think about how those tools fit together, who maintains them, and what happens when the people involved change. It is an engineer's concern with the joints in the structure.

He made a related case for tailored software that summer. Organizations have legitimate reasons to need something specific: an unusual operating process, a requirement, a way of working that a standard product does not accommodate. The problem is the expense and fragility that can accompany building everything from scratch.

His chosen analogy was buying a suit. Fit matters, and alterations have a purpose. A configurable platform could offer room for an organization's particular needs without demanding a wholly new foundation each time. For someone selling enterprise software, it was a commercial argument. For someone using it, it was a plea to consider the work before insisting on the wardrobe.

The same year, he attended BridgeSF and wrote about smart cities. Rather than stay with the spectacle of a connected metropolis, he used the logistics of getting children to football practice. The family carpool contained enough changing arrangements to make a dispatch office sympathize.

He identified trust and incentives as obstacles. Better coordination needed access to information people might hesitate to share. Public agencies and private companies also had different reasons for acting. Once again, the technical possibility was only part of the problem. People had to agree to participate, and the arrangement had to make sense to them.

Munich, with paperwork

By November 2018, Moustakas was describing a different sort of organizational puzzle. DevonWay was a 50-person American software company working on opening an office in Munich. He offered a candid account of the process while it was still underway, including the possibility that it could fail.

His advice started with the person making the move. Find someone who knew the organization, wanted to relocate, and had skills suited to the purpose of the new office. That person would represent the business. Enthusiasm and judgment belonged in the planning alongside location and cost.

Then came the less photogenic work: local expertise, company formation, banking, visas, and other administrative arrangements. His humor about unfamiliar paperwork made expansion sound like something people actually have to do, with a learning curve and plenty of small tasks. Geography changes faster in a presentation than it does in an operating company.

The account reveals an executive willing to describe work before it becomes a polished success story. At that point Munich was a process, with choices still to make. There is value in that tense. It leaves room for the effort, uncertainty, and dependence on other people's knowledge that a retrospective can tidy away.

“My single most important responsibility is people.”

Chris Moustakas · 2019

The people behind the usernames

In 2019, Moustakas wrote about a difficulty particular to enterprise software: users often arrive because a job requires it. They are trying to complete a task. They have little reason to linger admiring the application, and may have had little choice about buying it in the first place.

For the team making the software, that can obscure the value of the work. His answer was contact: user groups, visits, conversations with end users, and other exchanges that let customers and employees meet each other. A username could become a person with a recognizable problem. A feature could become something that helped that person.

His emphasis was on employees knowing that their effort mattered. That gives a different scale to the question of company culture. Appreciation has something concrete to attach to when the people building a product can see where it works. The feedback travels in both directions: customers get a human connection, and developers get a reason to care about the next improvement.

A tablet must earn its place

Later in 2019, he attended an Electric Power Research Institute conference in Charlotte about deploying mobile work management. His account concentrated on the practical measurements: how long documents took to load, whether tasks took the expected time, how much work happened digitally, and how reliably the application behaved.

One issue received particular attention: losing a user's work. If someone suspected a device could erase what they had done, confidence in the technology would suffer and paper would become attractive again. The insight is almost comically ordinary, which is why it matters. Paper has limitations, but a completed page usually remains a completed page.

In 2021, his argument for digital transformation followed the same careful rhythm. Regulated organizations accumulate processes over time. Change needs evidence, and employees need a manageable way to absorb it. He recommended starting with a process or two, setting measures of success, and expanding as results justified it.

That approach gives the user a part in the story. A rollout happens to people who already have work to finish. Introducing a smaller change lets them learn, react, and see whether it helps. It also lets the organization discover something before committing to more. Progress becomes a series of observable decisions, each providing information for the next.

A sale, then a new operating role

In September 2023, Ideagen announced an agreement to acquire DevonWay. Moustakas said further investment was needed for the next stage, and described a competitive process lasting several months. His explanation for choosing Ideagen included its culture, strategic direction, and customer base.

The transaction brought DevonWay's configurable operations software into a larger organization. Madison Park Group, which advised on the sale, subsequently confirmed the acquisition. Moustakas later served as Ideagen's senior vice president and general manager for North America.

September 2025 brought another move: chief operating officer at YuzeData, the sustainability data and intelligence business founded by Chris Joseph. The appointment connected his experience running enterprise software with a company focused on connecting information across systems. In explaining the move, he pointed to the value of bringing context to systems of record as AI developed.

The operator opens the terminal

Two months after the YuzeData announcement, he published an account of using Claude Code to understand business data. He described combining existing systems, retrieving information, and keeping snapshots so he could examine change over time. The ambition was to move from seeing what was happening to discussing what to do about it.

The project is a recognizable continuation of his earlier interests. Systems already contain useful information. An operator needs a view shaped around the questions at hand. Connecting the pieces can spare someone another excursion through several applications and another spreadsheet assembled for a single meeting.

He also described putting explicit instructions around the AI's behavior, including restrictions on invented numbers and deletions. Those details are telling. An executive experimenting with a new tool still has to decide what an acceptable answer looks like. A fluent response has limited value if its figures cannot be trusted.

In April 2026, his published account of YuzeData's AI approach set out commitments on customer data, third-party agreements, and visibility before deployment. It said customer data would not be used to train models. It also described more granular controls being built for organizations with particular governance needs.

The distinction between existing commitments and controls still being developed gives the work its proper shape. There are useful possibilities, and there are decisions to make about permission, accountability, and the conditions under which a tool operates. These are operating questions as much as engineering questions.

The jar of jellybeans belongs at the beginning because it makes discovery seem approachable. Moustakas's subsequent career supplies the less sugary part: observations need a dependable place to live, people need to understand the tools, and change needs a way to be judged. The prize, in enterprise software, is often simply being able to get on with the job.