Insurance coverage is an $8 trillion world trade burdened by handbook workflows and a rising expertise scarcity. Cara delivers an AI-native answer on AWS that automates back-office processes for insurance coverage brokerages.
Insurance coverage brokers routinely spend hours on repetitive duties. These embody finishing functions, analyzing coverage coverages, re-keying knowledge throughout programs, and relaying info between purchasers and carriers. Because the trade faces a persistent expertise scarcity, brokerages must scale income with out proportional headcount will increase.
On this submit, we discover how Cara, inbuilt cooperation with AWS, addresses these challenges. We stroll by way of the technical design choices and the AWS companies that assist the answer. We additionally share measurable outcomes Cara has delivered for enterprise brokerages.
The problem: Why generic AI falls brief in insurance coverage
Insurance coverage brokerages function in a extremely regulated atmosphere. Each transaction calls for precision, auditability, and compliance. The info concerned consists of delicate personally identifiable info (PII), monetary information, and underwriting particulars.
Generic AI instruments usually are not designed for this complexity. Efficient AI for insurance coverage should perceive domain-specific knowledge fashions and brokerage workflows. It should additionally deal with carrier-specific necessities and regulatory constraints whereas assembly enterprise safety requirements.
Cara’s founding workforce noticed these gaps firsthand. Vic Yeh, Nikhil Kansal, and Jon Patel beforehand based a digital insurance coverage brokerage. They scaled and offered it to The McGowan Firms, one of many largest privately held insurance coverage organizations within the US.
Throughout that have, the workforce constructed an inner AI copilot powered by giant language fashions (LLMs). The copilot diminished turnaround instances, improved knowledge accuracy, and streamlined agent workflows. Inspired by robust adoption, they expanded the idea right into a standalone product: Cara.
Structure overview
Cara is constructed on AWS companies chosen for reliability, scalability, and safety. Determine 1 reveals the high-level elements of Cara’s manufacturing deployment.
Cara structure on AWS
Compute and orchestration
Cara runs on Amazon Elastic Kubernetes Service (Amazon EKS) for container orchestration throughout a number of Availability Zones. EKS manages Cara’s microservices, together with ingestion pipelines, workflow engines, and the inference layer.
This structure helps elastic scaling to deal with demand throughout peak renewal and servicing intervals. It helps hundreds of concurrent customers and workflows per brokerage. Every group’s workloads run in remoted namespaces for tenant separation.
AI and inference
Cara’s AI capabilities are powered by LLMs hosted on Amazon Bedrock. Amazon Bedrock supplies entry to basis fashions by way of a totally managed API. This enables Cara to run inference with out managing GPU infrastructure. Cara makes use of Amazon Bedrock for a number of core capabilities:
- Protection and quote intelligence – compares provider quotes, summarizes protection variations, and highlights exclusions or gaps.
- Software and type automation – cross-fills ACORD and supplemental varieties utilizing supply paperwork, prior submissions, and company tips.
- Proposal and renewal era – produces branded, client-ready proposals and renewal spreadsheets.
- Data-driven workflows – references agency-specific tips, provider appetites, and historic placements to information choices.
Safety and knowledge isolation
Information safety is a foundational requirement for insurance coverage organizations. Cara’s structure makes use of account-specific deployments on AWS. Every brokerage’s knowledge and workflows are remoted inside devoted, safe workspaces. This design helps compliance with trade laws and supplies auditability on the group stage.
Integrations
Cara integrates with main company administration programs (AMS) and buyer relationship administration (CRM) instruments. It syncs accounts, insurance policies, and paperwork to scale back duplicate knowledge entry. AI-driven workflows function instantly inside current dealer expertise stacks. This design helps decrease modifications to the programs their brokers already use.
Deployment and operational traits
One in every of Cara’s design objectives is quick time-to-value. Enterprise brokerages can get onboarded inside hours and launch personalized workflows inside days. Cara’s deployment on EKS makes use of parameterized templates for every new tenant. It provisions remoted namespaces, storage, and inference endpoints with out handbook setup.
In manufacturing, Cara’s infrastructure on AWS supplies:
- Excessive availability – multi-AZ deployment on EKS with automated failover.
- Elastic scaling – Kubernetes Horizontal Pod Autoscaler adjusts capability primarily based on real-time demand. This helps hundreds of concurrent customers throughout peak intervals.
- Enterprise safety – knowledge isolation per tenant, encryption at relaxation and in transit, and integration with AWS Id and Entry Administration (AWS IAM).
Measurable outcomes
Cara’s AI-driven workflows have delivered quantifiable outcomes for enterprise insurance coverage brokerages:
| Metric | End result |
| Time saved per consumer | ~10 hours per week by way of workflow automation and contextual data retrieval |
| Onboarding pace | Enterprise brokerages onboarded inside hours; customized workflows dwell inside days |
| Concurrent capability | Hundreds of concurrent customers and workflows per brokerage |
| Adoption | Utilized by a whole lot of main insurance coverage companies and brokerages |
These outcomes come from organization-specific workflow automation and contextual data retrieval. They rely on Cara’s domain-specific AI and the scalable, safe infrastructure supplied by AWS.
Wanting forward
The insurance coverage trade stays within the early phases of AI adoption. As enterprise demand grows, Cara continues to broaden its AI-driven workflows throughout gross sales, servicing, and operations.
“We’re thrilled to advance the boundaries of domain-specific AI in real-world insurance coverage use instances with AWS,” says Vic Yeh, CEO of Cara. “Our purpose is to assist insurance coverage professionals return to the core of our trade: the relationships.”
Conclusion
On this submit, we confirmed how Cara constructed a domain-specific AI answer for insurance coverage brokerages utilizing Amazon EKS and Amazon Bedrock. The structure delivers tenant-isolated, elastically scaling workspaces. It helps hundreds of concurrent customers whereas assembly the safety and compliance necessities of the insurance coverage trade.
To study extra about constructing AI-powered functions on AWS, go to the AWS Structure Heart. To get began with Amazon Bedrock, see Getting began with Amazon Bedrock. For Amazon EKS, see Getting began with Amazon EKS.
In regards to the authors
