Cara Partners with AWS to Bring Domain-Specific AI to Insurance Brokerages

Cara is an AI platform built to automate the back-office work that insurance brokers spend hours on each week. The company was founded by Vic Yeh, Nikhil Kansal, and Jonathan Patel — operators who previously worked at Blend Labs, Stripe, and Strategy&. Before launching Cara, the founding team built and sold a digital insurance brokerage called Oyster Technologies to The McGowan Companies in May 2025. That exit gave them the on-the-ground experience to understand what brokerages really need: software that saves time on repetitive, high-friction tasks.

The platform uses machine learning to streamline workflows that traditionally consume enormous amounts of broker time. According to Cara's claims, the software can compress tasks that used to take 90 minutes down to just 2 minutes, and users report saving around 10 hours per week on average.

Cara has announced a partnership with Amazon Web Services, marking the company's first major public co-marketing effort with the cloud giant. The platform is built on Amazon EKS, which handles container orchestration, and Amazon Bedrock, AWS's service for working with large language models and AI inference.

The choice of these tools reflects Cara's focus on building what it calls "domain-specific AI" — AI systems trained and optimized specifically for insurance brokerage workflows rather than general-purpose AI. By using Bedrock, Cara can leverage multiple foundation models without building its own LLM from scratch, allowing the company to move faster and focus on solving the actual insurance problem.

The global insurance industry is worth roughly $8 trillion, making it one of the largest sectors in the world economy. Despite its size, much of the industry still relies on manual processes and legacy systems. Brokerages — the firms that help customers find and purchase insurance policies — are particularly affected by this inefficiency. Back-office work like policy management, compliance checks, and customer communication can eat up days each week.

Cara's timing is well-suited to the moment. The company closed an $8 million seed round on March 31, 2026, led by venture firm Kearny Jackson and supported by notable angels including Claire Hughes Johnson (former COO of Stripe), Kevin Mahaffey (founder of SNR), Sam Hodges (CEO of Vouch Insurance), and Colin Evans from OpenAI's startups team. The company has already reached seven-figure annual recurring revenue in just seven months, with 80 percent of that growth coming organically.

Insurance brokerages compete largely on service quality and responsiveness. A broker who can turn around client requests faster, manage more accounts without growing headcount, and reduce errors has a genuine competitive edge. Cara's approach suggests that AI, when built specifically for an industry's needs, can deliver real productivity gains rather than serving as a generic chatbot.

The AWS partnership also signals that the insurance industry is becoming a priority for cloud and AI vendors. As more brokerages move to cloud infrastructure and adopt AI tools, the pressure on smaller and mid-sized brokers to stay competitive will intensify. Those who adopt early may find themselves with a measurable advantage in client retention and growth.

Cara's reported metrics — the 10 hours per week saved, the compression of 90-minute tasks to 2 minutes, and the company's user scale — are self-reported figures published by Cara and relayed by AWS. These have not been independently audited by a third party. AWS, as Cara's partner and a vendor of the underlying tools, has a commercial interest in showcasing the platform's success. This is a standard partner spotlight, not an independent audit, so readers should view the performance claims as credible but not verified by outside sources.

Cara's success highlights a fundamental principle: domain-specific AI works best when it sits on top of clean, unified data. The platform can compress workflows and save time because it can access accurate, consistent information about policies, customers, and brokers — likely normalized from multiple source systems into a single source of truth.

For Australian mid-market businesses across insurance, professional services, finance, and other sectors, the lesson is clear. Before you invest in AI automation tools, ask yourself: where is my data coming from? Is it scattered across separate CRMs, accounting software, spreadsheets, and legacy systems? Or is it consolidated into one unified database that AI can actually learn from and act on?

Without that unified foundation, AI tools can only work with fragmented, incomplete information — which limits their effectiveness. With it, the same tools become genuinely transformative, just as they have for Cara's brokerages. The firms that build a single source of truth for their business data first will be best positioned to unlock the productivity gains that domain-specific AI promises.