At ALKEME, Ryan Deeds, Head of AI, argues that the insurance industry’s AI challenge is less about buying tools than building the culture, process discipline, and operating guardrails to use them well.
Insurance has no shortage of AI conversation. What it still lacks, in many organizations, is the operating environment that makes innovation usable.
After years inside agencies and agency networks, Deeds encountered that reality across several firms before joining ALKEME. ALKEME itself has grown into a large national brokerage platform, with over $2 billion in premiums written in 2025. But Deeds’ view is that scale does not remove the core problem.
For all the industry’s rhetoric around digital transformation, he suggests the real barrier is not technical readiness alone. It is whether an organization is willing to tolerate learning, variation, and some degree of controlled failure.
“I think if you have a culture that lives in fear, that is a culture that’s against innovation,” Deeds says. “They can say whatever they want, they can bring in whatever they want. The ROI of those new tools is going to be limited if everyone is afraid of making a mistake.”
In brokerages and agencies, innovation is frequently framed as a leadership priority, a systems issue, or a vendor decision. Deeds sees the adoption question as inseparable from day-to-day workplace dynamics, especially the divide between production and service.
In many firms, he argues, service teams bear the operational risk but do not always share the authority or cultural latitude to experiment. If mistakes are met with blame, hesitation becomes rational.
Deeds is not describing a future built only by central technology teams delivering polished tools from above. Increasingly, he finds, “Our success is going to be determined on allowing our users to create their own solutions in an environment that is secure, safe, and data centric,” he says.
That is a notable shift from the old model, where technology teams acted as gatekeepers and business users submitted requests into long development queues. Deeds’ view is that the spread of generative AI has changed employee expectations permanently. Workers entering the industry are less likely to accept that every small workflow issue requires formal IT intervention. If they hit a problem, many will try to solve it with whatever tools they can access.
The real governance challenge, then, is not stopping that impulse. It is channeling it.
Deeds is blunt about the alternative. If companies do not create approved pathways for experimentation, employees will build outside the system anyway.
Deeds’ answer is to begin by standardizing practices, embedding coding and data rules into internal workflows, and giving employees ways to test ideas without creating a “nightmare” of disconnected utilities and ungoverned tools.
Yet even that is only half the story.
“You have to have processes,” he says. “Show me an SOP that we’re following with an end result that we can test against. Sure, no problem. I can absolutely do that.”
That observation cuts against the current tendency to treat AI as a shortcut around operational mess. In reality, messy processes do not disappear when new technology is layered on top of them. If 15 people perform the same renewal process in 15 different ways, automation does not solve the inconsistency.
This is where Deeds offers one of the more useful distinctions in the current AI debate. A company may buy an “AI stack” and still see little return because it never established standard operating procedures, reliable outputs, or a shared understanding of what success looks like.
For Deeds, innovation is therefore less about chasing the newest model than building an organization capable of testing what is actually useful. “That to me is just being able to experiment,” he says. “That’s innovation today.”
Innovation tends to come not from abstract mandates issued at the top, but from the people closest to the friction. Account managers, especially, often see the problem more clearly than leadership because they are dealing with the real work rather than the version of the work described in strategy meetings.
“I always want to work close with the individuals and really understand what the pain points are,” he says.
That has shaped a build-fast, iterate-fast approach at ALKEME, where early wins matter because they prove the organization can actually solve something concrete.
There is also a broader strategic layer to this. ALKEME has expanded through partnerships and acquisitions, which means it operates across multiple offices, practices, and inherited ways of doing business. Deeds argues for a more selective model: introduce control where consistency genuinely matters, preserve flexibility where local differences are still productive, and use technology to create visibility without forcing every office into an identical mold.
Fast-growing brokerages often acquire strong firms precisely because they are winning in their markets. Replacing every operating habit on day one can destroy some of the value that made the acquisition attractive in the first place.
Still, Deeds does not underplay the stakes. Asked what advice he would offer insurance leaders trying to create a more innovative culture, he does not begin with software budgets or org charts. He begins with risk.
“I think they have to start with a conversation about how much risk is acceptable,” he says.
Insurance is an industry built to price, transfer, and control risk for clients. It is often less comfortable discussing the internal risk of standing still. Deeds’ point is that organizations need a more explicit view of both sides: the risk of moving too quickly, and the risk of waiting so long that employees innovate outside the enterprise or competitors pull further ahead.
At its core, the issue is not AI adoption. It is whether insurance firms can build institutions where experimentation is possible, process is clear, and trust exists at the working level. Until that happens, innovation will remain something many firms talk about fluently and practice only partially.
This article was produced in partnership with ALKEME
