YOUR VENDOR HIJACKED YOUR AI STRATEGY.

Michael Jones, Co-Owner and CEO, Combined Ratio Solutions
Luke Magnan, Co-Owner and Chief Insurance Officer, Combined Ratio Solutions

STOP APPLAUDING YOUR VENDOR’S AI STRATEGY

The insurance industry is drowning in AI theater. Every vendor, from the bloated legacy giants to the shiny-new insurtechs, has slapped AI on their pitch deck.

  • AI Submission Intake
  • AI Underwriting Assistants
  • AI Claims Triage
  • AI Fraud Detection

The press releases are relentless, the demo environments are polished, and not one of them is building AI for you. Stop applauding your vendor’s innovation for a second.

Ask the question they’re hoping you won’t:

“WHOSE AI STRATEGY ARE YOU ACTUALLY EXECUTING?”

These aren’t configuration settings you can dial up or down. They’re baked into the architecture. Locked behind the black box. And they were decided by someone whose incentives have nothing to do with your operational objectives.

THERE IS NO CHOICE

Here’s the con hiding in plain sight: your vendors are making consequential, opinionated decisions about how AI runs inside your business. Every model choice. Every automation threshold. Every data flow. Made without you. Announced to you after the fact, buried in a release note.

IT’S NOT INCOMPETENCE. IT’S A MODEL.

This isn’t malicious. It’s just how software gets built for the masses. Vendors serve the median customer. They optimize for the average underwriting philosophy, the generic claims culture, the one-size-fits-all risk appetite. Your differentiation, the thing that actually makes you competitive, doesn’t factor in.

YOU BOUGHT CONVENIENCE… NOT INNOVATIVE TECHNOLOGY

The insurance vendor AI strategy being deployed today is the same approach that we have seen over the last 10 to 15 years. It’s a playbook the insurance tech industry runs on repeat: dominate one domain, then expand everywhere else. Call it a suite. Sell it as convenience. Watch your customers get trapped.

Deep expertise doesn’t clone itself across domains. When a vendor goes full-suite, they’re trading mastery for coverage. Depth for breadth. And now they’re layering AI on top of that alreadycompromised foundation. The tradeoffs don’t disappear. They stack.

The result: you get claims-native AI trying to think like an underwriter. Billing logic governing policy administration intelligence. The AI knows what the vendor knows, not what you need.

The AI world outside insurance is moving at a pace no vendor can match. Foundation models today are unrecognizable compared to eighteen months ago.

Eighteen months from now, today’s capabilities will look like dial-up internet.

Every decision your vendor makes, which models to use, how often to update them, what use cases to prioritize, what data to expose, is a decision about how fast you can move. They control your clock. Every quarter inside their closed AI ecosystem is a quarter your more nimble competitors spend pulling ahead.

THE MCP ADVANTAGE

(Model Context Protocol)

The mechanism that makes open architecture practical, not just theoretical, is MCP, an open standard now governed by the Linux Foundation. CRS OSPolicy implements MCP as the integration layer between your policy data and any AI system you choose to deploy.

Implement once: MCP is an open standard now under the Linux Foundation. Connect to any AI swap models as the market evolves; no integration rebuild.
Connect to any AI: Claude, GPT, the model that launched last quarter and beats both of them. You’re not locked to whatever AI vendor your policy admin provider struck a deal with.
Turn it off anywhere: The AI module is optional. Governance not ready? Regulation shifted? Off. No support tickets. No waiting for a release cycle.
No vendor lock-in: Model-agnostic by design, open by architecture. Your AI spend follows your competitive strategy, not your vendor’s product roadmap.

Controlled AI. Not Blind AI Adoption.

This isn’t a pitch for AI maximalism. CRS uses AI deliberately in our own delivery to accelerate insurance product creation, reduce repetitive work, and speed iteration. But senior architects and developers own the judgment, the design decisions, and accountability. That’s the operating model we're enabling for clients: AI as a force multiplier inside a governed process, not a substitute for one.

VENDOR LOCK-IN ISN’T NEW

But in the AI era, it’s more dangerous than it’s ever been, because the stakes are no longer just switching costs. They’re strategic velocity. Every quarter inside a closed AI ecosystem is a quarter your more nimble competitors spend pulling ahead. The bottleneck isn’t the technology. It’s who controls it. Closed systems mean the vendor sets your risk tolerance, governs your data flows, and decides which AI capabilities you’re allowed to access. You don’t get to audit what you can’t see.

When insurers own their AI strategy, the difference is immediate and concrete.

 
You pick your models. Claude, GPT, the model that launched last quarter and beats both of them, you’re not locked to whatever AI vendor your policy admin provider struck a deal with.

You set the priorities. Speed of quote? Underwriting sophistication in a niche class? Your AI spend follows your competitive strategy, not your vendor’s product roadmap.

You move when you’re ready. Better model available? Evaluate it. Regulation shifted? Turn that module off. No support tickets. No waiting for a release cycle.

You stay in control of the outcome. We use AI ourselves, to accelerate insurance product creation, reduce repetitive work, and speed up iteration. But senior architects and developers still own the judgment, the decisions.

“ARE YOU BUILDING AI FOR US?”

The only question to ask. In every renewal conversation. Every vendor evaluation. Every platform expansion pitch.

The answer tells you exactly what relationship you’re signing up for and exactly what strategic position you’ll be defending, or surrendering, for the next decade.

A Reality Check: Where Do You Actually Stand?

The carriers who define the next era won’t be the ones who picked the best vendor AI subscription. They’ll be the ones who built real data assets, made deliberate model choices, and had the platform freedom to act on what they learned.

Policy administration should be the foundation of your AI strategy.

Not the ceiling of it.

 

Ask your team these questions:

  • Can you deploy new AI capabilities or test new models without vendor approval?
  • Can you adjust underwriting workflows or automation rules on your timeline?
  • Can you access, export, or use your own policy data without restriction?
  • Can you integrate a new third-party capability without waiting for a release cycle?
  • Can you turn off AI in any specific workflow without a support ticket, a release cycle, or a contract amendment?

If any of those answers is no, your AI strategy isn’t yours. And deep down, you already know it.

LIBERATE YOUR AI STRATEGY

If you’re done inheriting your vendor’s AI roadmap, their model choices, and their release cadence, let’s talk. We’ll map where vendor lock-in is costing you speed, and what an open, MCP-based policy administration foundation would change.
 
 

Control It. Own It.

combinedratio.com

GOT A TECH CHALLENGE?

We’ve got the experts whether it’s seamless implementation, no-drama integrations, custom builds that actually work, or support that doesn’t ghost you.

Let’s make your tech better.

DEVELOPER RESOURCES

Source Code (GitHub)

Knowledge Base

Open-Source Forum

Start Your Rebellion

To gain access to the open source code, submit your information, and a Rebel leader will be in touch.

    Contact your Rebel

    Please use a business email address or email us directly at rebels@combinedratio.com