(Alec Miloslavsky, CEO, EIS. Photo by author.)
When insurance executives talk about generative AI, Alec Miloslavsky, CEO, EIS (San Francisco) detects a familiar mix of enthusiasm and uncertainty. Leaders recognize the technology’s immense efficiency potential, yet many lack a coherent framework for how to approach it. The bigger concern, he suggests, is not whether AI will matter—but how to embrace it without becoming dependent on a rapidly consolidating technology industry whose business model favors “winner-take-most” economics.
“If I were sitting in an executive chair of an insurance company,” Miloslavsky explains, “I’d be thinking about three or four things at a high level—where this is going, how to get ahead of it, and how not to become beholden to the tech industry for what I’m doing.”
This tension—between opportunity and dependency—frames EIS’s evolving strategy around AI.
From Features to Transformation
Miloslavsky categorizes the company’s approach to AI into three distinct tiers of increasing strategic importance.
First are narrow, targeted capabilities embedded within specific business functions. Use cases such as rating analysis, claims intake, and fraud detection already leverage generative AI behind the scenes. To the user, however, these applications appear as packaged features rather than experimental AI tools. Because they are contained in scope and measurable in impact, they lend themselves to traditional ROI evaluation while minimizing risk.
“Because it’s narrow, it’s contained,” Miloslavsky notes. “You can judge the relationship between the price and the benefit easily.”
The second area focuses on accelerated time-to-market through architecture. EIS has embedded generative AI directly into its development tools, indexing documentation, historical deployments, and internal knowledge repositories. As a result, business analysts can describe system changes in natural language, with AI guiding the configuration process—often eliminating traditional development cycles entirely.
“It’s doing the same thing, only faster,” he says. “In some cases eliminating a technology deployment loop.”
This capability shifts power from technical teams alone to a broader group of business stakeholders, allowing insurers to respond more quickly to market demands.
The third and most transformative area is full agentic orchestration.
By exposing more than 12,000 APIs through an agentic interface, EIS makes it possible for users to issue complex, multi-step instructions in plain language. A request such as, “Archive all customers aged 26 to 30,” can be interpreted, orchestrated, and executed by the platform across thousands of processes—without manual coding.
“You can literally talk to our system in plain language,” Miloslavsky says, “and the system can orchestrate what you’re asking.”
Guardrails for a Powerful New Interface
This degree of flexibility brings significant risk, and Miloslavsky is direct about the need for governance.
“What could possibly go wrong?” he jokes—before detailing the serious implications that free-form access could have for security, compliance, cost management, and system stability. Every AI instruction carries computational cost and regulatory consequences if left ungoverned.
EIS addresses this through:
- Role-based authorizations and activity controls
- Continuous usage monitoring and cost governance
- Alignment with emerging regulatory standards
- Pursuit of available certifications for responsible AI use
- Participation in global forums and open-source communities focused on AI governance
“There is no zero-risk path,” Miloslavsky acknowledges. “You’re choosing the path that minimizes risk, not eliminates it.”
In that context, EIS increasingly acts not only as a technology provider, but as an intermediary—helping insurers balance innovation with compliance, safety, and reputational protection.
Preserving Optionality Through Agnosticism
Another cornerstone of EIS’s philosophy is optionality. The company designed its platform to remain both cloud-agnostic and model-agnostic, avoiding lock-in to a single provider such as OpenAI or Amazon Bedrock.
“It’s not just pricing,” Miloslavsky observes. “It’s intellectual property and ultimately control.”
By ensuring its AI implementations are portable across vendors, EIS protects both itself and its customers from future market consolidation or strategic vulnerability. If a dominant provider falters—or gains excessive leverage—customers retain flexibility to shift without dismantling core operations.
“This is risk management,” he says plainly.
Architecture as a Catalyst
Miloslavsky points to EIS’s modular, API-driven cloud architecture—referenced in the industry as MACH (Modular, API-driven, Cloud-native and Headless) model—as a key differentiator. EIS was the first core insurance platform certified by the MACH Alliance, he notes. The platform is architected with interoperable components designed from inception for connectivity, extensibility, and scale. Miloslavsky contends that this flexible, API-native, cloud-first origin has allowed the company, and its customers, the flexibility to advance more rapidly than competitors still transitioning legacy systems.
“They’re still fighting to get into a smooth transition into the cloud,” Miloslavsky comments. “We never had to do that. That’s why we can say we are AI-ready.”
This architectural advantage now extends into AI adaptability, allowing EIS to incorporate new capabilities without major structural rework.
The Real Issue: Adoption
For Miloslavsky, the ultimate challenge is not technical—it is organizational. He likens AI’s emergence to the rise of Java: an unavoidable utility that becomes part of the underlying fabric of how work is done.
“You can’t not use this technology,” he says. “It has to be your basic platform.”
EIS’s strategy, therefore, is to encourage widespread, responsible engagement by embedding AI as a native capability. Anyone in an organization can interact with it—bringing in third-party tools, leveraging built-in functions, and working across departments in ways that were previously constrained by technical barriers.
“Our job is to foster that adoption,” Miloslavsky explains, “and bring the know-how on how to do it responsibly.”
In a landscape defined by rapid change and unresolved questions, EIS is positioning itself not merely as a solution provider, but as a guide—architecting for agility while anchoring innovation in discipline, governance, and long-term strategic control. For carriers preparing for a world of agentic AI, Miloslavsky suggests, the winners will be those who not only deploy AI, but adopt platforms structurally designed for it—platforms capable of orchestrating intent, managing risk, and keeping options open as the technology accelerates.