Phil Verghis
Founder, Verghis Group
University of New Hampshire, early 1990s
My first job out of engineering school, in the early 1990s, was at the University of New Hampshire, doing tech support. I’d realized I liked working with people on actual problems, and tech support was where they came for help. I’d come to technology late, and once I saw what it could solve I was obsessed with it. From the start I was trying to build one of the first AI-powered knowledge bases to handle the questions that kept coming back. The technique was called case-based reasoning — one of the early AI approaches, where the system learned by matching new problems against the shape of problems it had already seen solved. What I learned in the process was that the technology wasn’t the hard part. The people and the process were.
I scoured for tools and training on how to set up a help desk well. There was nothing — no books, no courses, no community of practitioners I could learn from. This was before the web was something you could search, before any of it was written down anywhere a junior engineer at a small state school could find. So I wrote what I needed myself, and I shared everything I learned with anyone who asked. Eventually I put it on BitNet — the predecessor to the internet, the network most universities ran on at the time. By the time I stopped updating it, over a decade later, it had spread to 103 countries.
By the mid-1990s we’d also stood up one of the earliest online help desks at UNH.
Duke University, late 1990s
Duke was where the work got serious. The UNH CIO — my boss’s boss — had moved to Duke to lead IT across the campus and the medical center. The two organizations hadn’t integrated, but their leadership had decided to collaborate closely from the top down at multiple levels. She asked me to join the effort to figure out a new way of working.
This was one of the earliest cross-functional efforts of its kind in higher education. The medical side ran in real time — the right information had to reach the right person at the moment they needed it. We used to say it bluntly: babies could die. The campus side ran on different rhythms, built around faculty, staff, and student support.
The brief from the CIO was simple — work together, make it better. I had ideas. I led the build of two of them: DUNK, the Duke University Networked Knowledgebase, holding the institutional memory of how the work actually got done, with a partner from the medical center. And the live entry point both sides could use — taking what I’d built at UNH and rebuilding it for Duke as one of the earliest web-based help desk systems anywhere. The help desk system won Network World’s Service Excellence Award. DUNK won its own recognitions, including a PC Week photoshoot that involved me dunking on Coach K’s court with a strategically placed stepladder.
Akamai, 1999
A colleague from Duke had been turning down offers all through the dot-com boom — he wasn’t going to leave Duke. Then he did. He joined a small MIT spin-off nobody had heard of, called Akamai. A few months later he called and asked me to join him. I did. The IPO that October was one of the largest of the era — heady, for someone fresh out of academia.
Akamai was a new category of infrastructure. Nobody had built it. Nobody knew how. Customers arrived faster than we could onboard them, and we were inventing the technology and the org while we ran them. I was completely out of my depth — so was everyone — and the fun was that we were surrounded by very smart people, figuring it out together. I sat at the intersection of customer-facing work, the internal company, and operations — because that’s where the actual work was.
When 9/11 happened, we lost Danny Lewin — one of our co-founders, the mathematical mind behind the algorithms — on the first plane hijacked, before it hit the World Trade Center. The Founder’s Award I’d won the year before was renamed the Danny Lewin Award in his honor. In the weeks that followed, with the company under pressure and the telco networks we ran on collapsing, I was asked to take on Global Network, Operations, and IT on top of Service Delivery. The company needed someone who could work well across — who understood the different groups, and knew how to take care of customers and partners as well as the infrastructure. By then I was starting to resemble a person like that, so they took a chance on me. I stayed through the dark days, helping the company shift its customer base from dot-com to enterprise and grow the company to profitability as part of the extended leadership team.
It was a profound crucible. What I came out of those years with — the relationships, the learnings, a way of seeing the work — I’ve carried with me. Many of those colleagues have gone on to become CEOs of their own companies.
On my own, 2004
By January 2004, Akamai was stable, profitable, and on a path to being the multi-billion-dollar infrastructure platform it is now. I left to start the first version of Verghis Group. I wanted to see what I could do without that kind of safety net underneath me.
What I was thinking about in those years was the same thing I’d been working on at Duke and Akamai: what happens to work when it crosses between groups, and what changes when those groups own it together. I wrote about it. I spoke about it. The Ultimate Customer Support Executive came out. IBM commissioned a report based on it, designated me an IBM Thought Leader, and flew me around the world to talk about it. Engagements followed — Fortune 500s, mid-market, PE, startups — from the IBM talks, from other conferences and workshops, from companies doing similar interesting work.
Along the way I chaired HDI’s Strategic Advisory Board — one of the apex associations in the industry — and had the chance to play a pivotal role with the founding teams of two companies, both still significant enterprises in the field today.
Klever Insight, 2013
In 2013 I closed the first version of Verghis Group and co-founded Klever Insight. The goal was software for service and support teams — to help them share what they knew, solve problems faster, and lift the operation as they went.
Our first two paying enterprise customers were UCOP and HPE, joined later by Salesforce. UCOP — which runs operations across the ten University of California campuses, Berkeley and UCLA among them — came in through a former Verghis Group client who’d moved over as a senior leader. HPE came in from a long relationship that ran back to those earlier consulting years. Both shaped the algorithms underneath what we were building.
What went into the software was everything that would have powered it — the algorithms, the principles, the way of seeing the work. Early versions of ServiceDNA, the instrument for reading the conditions underneath an operation. Time to Smile, the way of measuring how fast value lands — for the customer, for the team. With peers in the field, we built a cross-functional way to measure customer outcomes — the Open Customer Metrics Framework. Underneath all of it: bringing people along — guiding, not grading.
Klever ended in 2023 — didn’t quite survive COVID. The software never shipped. Many of the principles have stayed.
I’m grateful to the team who joined me, the investors, the board of advisors, and the early customers who took a chance on what we were building.
During this period I joined the boards of advisors at TSIA, Support Driven, and TSANet — three of the most influential associations in the field. (TSANet is the cross-vendor alliance member companies turn to when a customer issue spans more than one of them.)
LuxCreo, 2023
Mike, one of my friends and colleagues from the Akamai days, had become co-founder and CEO of LuxCreo — the 3D-printing platform behind medical-grade dental devices for same-day chairside printing of aligners, retainers, night guards, and surgical guides. Work that used to take a dental lab weeks. He brought me in as a consultant first; a few months later I joined full-time as inaugural Chief Customer Experience Officer.
The role gave me something I couldn’t have gotten as a consultant: accountability for the services operations. The work spanned hardware, AI software, materials, and global logistics — pulling dental manufacturing out of specialized labs and into the dental practice itself, where it could happen chairside in a single appointment. We pulled every group together — engineering, manufacturing, logistics, sales, IT, support — around a single end-to-end measure: Time to Smile, from the moment the customer placed the order to the moment they were printing on their own. Every group held to the same outcome.
We chose DevRev for the platform underneath all of that — an operations platform built around AI from the ground up, for service and customer-facing work, rather than retrofitted onto older architecture. It supported the work across every group — including the reseller-ready portal we needed to take the support operation from consumer-grade to enterprise-ready.
Now, 2025 onwards
Right now I’m working with leaders thinking through what AI is about to do to the operations they’ve built.
The question is urgent in different ways depending on who’s asking. A PE firm wants what they learn at one portfolio company to travel to the others. A Series C company in explosive growth needs the AI work to land inside an operation being rebuilt every six months. A leadership team that’s already deployed something and watched it not land the way the pilot promised wants to know what they missed.
I’m on the board of advisors at DevRev — the AI-native operations unicorn.
The clients I do my best work with are interesting people doing interesting things — who care about their customers and their employees in the same breath, who push and let themselves be pushed, who want what they’re building to be better for everybody it touches, and who give back in whatever form their work makes possible.
If any of that sounds like the seat you’re sitting in, let’s have a conversation. It might be interesting for both of us.
The first conversation is thirty minutes. Tell me what you’re working on.