When Halwell hired technologist Stephen Matusiak (pictured) and told him to “fix everything,” it set in motion a decade‑long shift away from vendor dependence and paper‑driven processes towards structured data, automation and in‑house ownership of critical systems.
When Stephen Matusiak arrived at Halwell l just over a decade ago, the brief from the chief executive was disarmingly simple: fix everything.
“Basically, everything was broken,” he recalled. “Nobody was happy with their technology. Everybody was grinding their gears and really upset with the way that things were.” The company was wrestling with creaking systems, paper‑heavy processes and a heavy dependence on external providers. What it didn’t have was a single person on staff who both understood technology and was accountable for it.
That, in Matusiak’s view, was the decisive turning point.
“Our CEO made what I think was a really good decision, which was: you need somebody on staff who is responsible for technology,” he said. “A lot of small companies outsource all of their technology, and they do not have anybody who is actually a paid employee who is responsible for the technology or who understands the technology.” That philosophy has become a key theme in his advisory work with other organizations: outsource elements of support if you wish but never outsource accountability.
From that starting point as a lone systems administrator, Matusiak has since risen to the CIO role, reshaping not only the insurer’s technology stack but also its culture, operating model and relationship with brokers.
He describes himself unapologetically as “a computer guy or a technology guy, not so much an insurance guy”. He studied Information Systems at St. Francis Xavier University in Nova Scotia on a football scholarship, only to see the scholarship evaporate after his first‑year grades collapsed under the pressure of dual commitments. Losing that support, he said, was “pretty upsetting”, but it forced a reset: he dropped the sport, focused on study and ultimately graduated at the top of his class.
From there, he followed his partner’s academic postings: first to St. John’s, Newfoundland, then back to Ontario for her PhD. The move to Newfoundland left him in a city with “not a booming tech scene,” and led to months of unsuccessful applications before a role emerged in university research IT. That position came with a significant upside. It gave him direct responsibility for a “quarter‑million‑dollar enterprise class server” underpinning genetic research which became the springboard for a series of infrastructure and operation roles once the couple returned to Ontario.
Subsequent roles saw him working on help desks, data centres and email environments, always close to the plumbing of systems. By the time Halwell Mutual came calling, he was accustomed to running critical platforms rather than simply consuming them. From IT Manager at Halwell he was soon promoted to VP of Business Innovation at the age of 30. In 2024 Halwell Mutual and Dumfries Mutual amalgamated to make HD Mutual. Matusiak at this point took on the role of CIO.
Halwell was also a business that had locked itself into the comfort and constraints of a monolithic approach. Most technology was outsourced to a managed service provider (MSP). The internal view was shaped by a mix of fear and dependence.
“I find technology is scary to a lot of people, and I think the biggest reason for that is because people do not understand it,” he said. Executives knew they needed to “take advantage of it,” but lacked the confidence to challenge external experts. The result, seen often in the sector, was over‑reliance on a single vendor and an all‑purpose platform: “we have this one platform, and this one platform does some things well and some things really poorly, but we try to jam everything into that one platform.”
That higher value work began with a forensic look at how people were actually spending their days.
He described walking into an underwriting manager’s office to find two towering stacks of paper and a ruler on the desk. One stack was a regulatory report; the other, a printout of all system transactions for the month. The plan was manual reconciliation, line by line, to find any missing entries.
“When I found somebody doing something that I would really hate doing myself, I stopped and said, ‘Let me help you with this. Let me figure out how to fix this problem for you,’” he said. In some cases, Excel and basic data handling were enough. Elsewhere, more structural changes were required.
Underwriting workflow was one such case. All broker communications, new business, endorsements, cancellations and queries, arrived in a single shared email folder with “thousands of unread emails” and no way of knowing what lay beneath. There was no breakdown by type of work, complexity or effort. With a background in IT operations, Matusiak recognized a pattern he knew how to solve.
“Working in IT, I have always worked in ticketing systems. Ticketing systems let you manage and organize all your backlog,” he noted. The answer, he decided, was to move away from a raw inbox and implement a proper ticketing solution. He implemented Zendesk himself, working closely with underwriters to ensure it matched reality on the ground.
He regards that project as a watershed: the moment he stopped simply “managing the technology” and began “finding new solutions to solve real business problems.”
The shift from unstructured to structured data sits at the heart of his strategy. Emails and documents, left as they are, do not lend themselves to automation or analytics. When they are turned into discrete items, tickets, records, classified content with clear ownership and metadata they become raw material for improvement.
“Unstructured data is just an email that comes in or a document. Structured data would be taking that email and turning it into a ticket,” he said. “You need that structured data if you want to start automating anything in your business. Automation is the underpinnings of where you get the benefit of technology.”
On top of that, the insurer has become sharply selective about the tools it will consider. Matusiak looks for “software that is developer‑friendly, where it is highly customizable and has a well‑published and well‑supported API so that we can connect different systems to each other to talk.” The litmus test is simple: “if the vendor cannot hand us their API documentation and say, ‘This is what you can do programmatically with our software,’ we are not interested in it.”
What he has not done is redesign his brokers’ world. For distribution partners, the unit of work remains the email. Some brokers push out documents from their broker management systems, others write directly in their inbox, but the insurer is careful not to ask them to change how they operate.
“Our mandate has always been: we do not want to make it harder for the brokers. We do not want to break their current workflows,” Matusiak said. “We take whatever it is that they give us and then we try to make the best of it.”
Rather than pushing portals or new front ends, the company has focused on processing and structuring what arrives, preserving broker convenience while extracting efficiency on its own side.
Matusiak is also skeptical of the default instinct to outsource major implementations. His practice has been to acquire the software, explore it hands‑on, and build internal capability before turning to external specialists.
“With software implementations and things like that, my first approach is: get it and just play with it. See what it is capable of doing, see how hard it is to build different pieces of it, and then slowly ramp up,” he said.
“Every project that we have done where we have outsourced the implementation has had a shorter shelf life, has been more expensive for us, and has operated less well. We have found the most success by taking full ownership and accountability for the implementation.”
Owning the work has a further advantage: after going live, tweaks and enhancements are not contingent on fresh budgets or change requests. “Because we own it, we did the implementation, and we trained our own staff on how to manage these things, it is easier for us to maintain it over time. We are quicker to implement fixes along the way so that those inefficiencies do not pile back up,” he said.
On internal change, he is acutely aware of the human cost. Technology‑driven transformation often triggers deep anxiety among staff who feel their hard-won expertise is being devalued.
“I completely understand that feeling of, ‘I used to be an expert in what I was doing, and once you change everything, I am no longer going to be an expert,’” he said. “That is probably a really bad feeling for people to have.”
The insurer’s response has been twofold. First, it targets the tasks everyone dislikes most, the “constant re‑keying” and administrative grind, so that automation is clearly a benefit rather than a threat. Second, it has hardwired into its operating model an expectation that new processes should not add manual burdens.
“We do not implement manual processes anymore,” Matusiak said. “Anything new that we are expecting people to do should be heavily automated. If somebody has to log in and do something manually once a month, my answer is, no, we are not doing that.”
He also makes a point of describing what jobs will look like after change. In underwriting, that means less time entering data and more time on portfolio management, product design and pricing. That explicit picture of the future, he believes, “does a lot to settle the nerves for people who are worried that they are going to be out of a job.”
Perhaps the clearest endorsement of the approach is cultural. Engagement surveys and feedback exercises now show employees “really looking forward to automation.” As he puts it, “we now have a solid culture of people who are the ones asking for it. They want to see more automation. They identify and notice the places that could be improved by more effective use of technology.”
Artificial intelligence has become an important part of how our DevOps team at HD approaches both development and problem-solving,” he said. “We use AI not only to assist with writing code and troubleshooting, but also as another powerful integration tool within our technology stack.
“Our teams leverage capabilities such as Microsoft Azure Cognitive Services and a range of machine learning models for tasks including document classification, optical character recognition (OCR), entity extraction, and data cleanup. These technologies allow us to automate processes and solve challenges that were previously difficult, if not impossible to address through traditional programming approaches alone.
“AI has been on our radar at HD long before the recent surge in attention following ChatGPT. As early as 2018, we were experimenting with machine learning models to help structure our data and identify data integrity issues, laying the groundwork for more advanced capabilities.
“From my perspective, AI represents a natural evolution of digital technology. However, organizations should focus first on building strong digital maturity rather than rushing headfirst into the AI wave. When systems are well integrated and data is properly structured, AI becomes relatively straightforward to implement and scale. Without those foundations, the most organizations can realistically achieve is an intelligent chatbot rather than truly transformative capabilities.”
For an industry still grappling with legacy systems and cautious attitudes to change, the lesson from Matusiak’s journey is stark. The real transformation did not begin with a multimillion‑dollar platform replacement or a consulting‑led roadmap. It began when a mutual insurer decided to own its technology destiny and hired someone whose job description, at least to begin with, was simply to fix everything.
“Outside of work, I prioritize time with my family,” he said. “My partner, our daughter, and our two elderly rescue dogs. We enjoy long walks, trips to the park, and spending time outdoors whenever we can. Being in nature is one of the ways I disconnect and recharge.
“I also enjoy coding and tackling technical challenges. Building things has always been a passion of mine, and I’m currently working on an AI startup focused on digital sovereignty. The goal is to create an affordable “AI-in-a-box” solution that allows organizations operating in privacy-sensitive environments to benefit from AI without concerns around data leakage or data residency.”
In addition, Matusiak finds time to coach and mentor, helping both individual contributors and executive leaders strengthen their digital literacy and better understand emerging technologies.
