Recently I hosted a webinar with Jamie Henry, VP of Product at Origami Risk, who sponsored the session. We called it “From Systems of Record to Systems of Intelligence: The Next Evolution of RMIS.” (I noted on air that we need shorter titles. Still true.)
Jamie and I have these conversations over coffee anyway, so we did it in front of an audience, with data from this year’s RMIS Report survey: over 1,000 responses from risk and claims professionals who use these systems. We also polled the webinar audience live.
Both sources told the same story. There’s a wide gap between what these platforms can do and what organizations actually use them for. And the reasons are organizational, almost every one of them.
Everyone Uses a Third of the System
Claims management, claims administration, and incident intake dominate usage in the survey. No surprise. That’s what these systems were built for, and they do it well.
The drop-off after that should bother you. Renewal management, policy and program management, exposure data management: each used by roughly a quarter of respondents. These are core functions of risk. Nearly every organization does them. They’re just doing them in spreadsheets, sitting next to a system they already pay for that handles all three.
Jamie sees the same thing in Origami’s own usage data. Clients have access to these capabilities automatically, and they sit idle.
One bright spot: EHS usage is at 20% and GRC keeps climbing. Those functions historically lived adjacent to risk and claims, often in separate tools, so watching organizations pull them into one platform is real progress. That needle has moved for a couple of years running.
“We Know We’re Competing With the Spreadsheet”
That’s how Jamie challenges his product team, and it’s the most honest thing I’ve heard a product executive say in a while.
Because the spreadsheet usually wins on a technicality: the process isn’t broken. The values collection runs on email and gets transcribed into Excel, the renewal exhibit gets built by hand, and it all technically works. Moving any of it into the RMIS is a special project, and risk teams are resource-constrained. So the project never gets prioritized.
The cost shows up later. Disparate data, transcription errors, and decisions that are less data-driven than anyone would admit, because the information needed to answer a board-level question lives in 6 different files.
Our live poll confirmed the diagnosis. The top answer for what limits strategic value: not enough time to fully configure and adopt. Integration came right behind it.
The Shelf Problem
Here’s the stat pairing that defines this market right now. 72% of survey respondents aren’t using AI-driven capabilities. 47% aren’t using advanced analytics. 40% aren’t using workflow automation. And when we asked about priorities for the next 1 to 2 years, analytics and dashboards topped the list at over 50%, with reducing administrative burden and automated workflows close behind.
They want it. They believe they need it. They haven’t done it.
(AI itself sat lower on the priority list, near 28%, and I read that as healthy. People are naming the outcomes they want. AI is one of the tools that gets them there.)
I see the mechanism constantly in our consulting work. Organizations implement a RMIS, and then they put it on a shelf. The phase 2 list (“we’ll add these 3 things once we’ve got adoption”) never gets revisited.
New features ship and no one looks at them. The system keeps doing what it did on go-live day while the organization changes around it.
Switching platforms doesn’t fix this, and most respondents have tested that theory: the majority are on their second RMIS, some on their third. Jamie’s observation matches mine here.
Teams that switch often recreate exactly what they had, sometimes replicating manual processes inside a brand new system. The outcomes don’t change, the frustration returns, and the punch list goes back on the shelf.
Start AI Before Your Data Is Perfect
The AI numbers deserve their own look. Roughly 60% of organizations aren’t using the AI capabilities in their RMIS. A bit over 60% are using AI tools somewhere in their organization. People found the value; they found it outside their risk systems.
(The survey ran early this year. I’d bet money those numbers swing hard by spring.)
The biggest misconception Jamie hears, and one we each started with ourselves: that you need models trained on your own data, or on industry data, before AI is useful. Origami tested that assumption and found the foundational models are already powerful and accurate. Your data rides alongside as a protected sidecar: it makes the results sharper and more actionable, without the heavy investment of custom model training.
That doesn’t let anyone off the hook on data quality. Feed bad data in, get bad results out. And I’ll repeat something I told a client recently: when your location data is stale and the AI reports patterns based on it, the model did its job. The data lied. Calling that a hallucination misses where the problem actually lives.
An audience member asked whether they should hold off on AI until their data is clean. Jamie’s answer, which I endorse fully: no. Start now.
Two reasons. Your team needs to build the habit, and that only comes from use. And AI is remarkably good at telling you which data needs cleaning and in what priority order.
We just did this for a client on cause codes. Instead of trusting a years-old standard mapping, we ran the original carrier data to see which codes actually get used, then rebuilt the code list around their organization. That kind of ongoing data monitoring used to be nobody’s job. With AI it’s cheap enough to do continuously.
What I’d Do With This
Start from the outputs, then work backwards. Jamie’s advice on any implementation, re-implementation, or new module: figure out what reports the board and executives need, and the required data elements and process fall out of that. Choose the tool path last. Bringing your old process into a new system, unexamined, is how you end up re-buying your current frustrations.
Pull the list off the shelf. Find the phase 2 items you deferred at go-live. Then look at what your vendor has released since the last time you made any real changes. Some of it will move the needle for you now.
Move one spreadsheet into the system. My candidate is exposure and values collection, the underused capability that surprises me most. If policies, exposures, and renewal data live outside the RMIS, you can’t produce total cost of risk from the system. That analysis alone justifies the effort.
Add denominator data. Jamie’s healthcare example stuck with me: raw claim counts across facilities are apples and oranges until you bring in patient beds or employee counts per location. Whatever your industry’s denominator is, integrate it. Counts become rates, and rates tell you where to act.
Give AI one real task this quarter. Summarization of claim notes is the low-hanging fruit. Data profiling is the sleeper. Pick one, keep a human reviewing, and expand from what works.
The Limiter Has Changed
Five or ten years ago you could fairly blame the technology, and I’ve had plenty of arguments about whether a RMIS is an operational tool or a strategic one. (It should be both.) Today the capabilities outrun the execution.
The integration exists, the automation exists, and the AI is arriving fast, with vendors like Origami under real pressure to keep lowering the barriers to using it all. What separates the organizations getting intelligence from their systems, in Jamie’s words, is a dedication to continuous improvement and someone on the team willing to act as a change agent.
Neither of those requires a bigger budget. One problem at a time is enough.
The full RMIS Report is available on our website, and the webinar replay is worth your hour if this gap sounds like your organization and it probably does. Which means the advantage goes to whoever closes it first.
Patrick O’Neill is President and Founder of Redhand Advisors, the independent advisory firm focused exclusively on the RMIS and RiskTech market, and publisher of the annual RMIS Report.