InsurTech is currently witnessing a tectonic shift as legacy systems meet the “reasoning layer” of artificial intelligence. One of the primary architects of this change is General Magic, a firm that recently secured $7.2 million in seed funding to tackle the persistent friction in insurance distribution. By moving beyond simple chatbots to deploy sophisticated AI agents, the company is shortening quote times from half an hour to just minutes, fundamentally altering how carriers and brokers interact with their policyholders.
This interview explores the mechanics of accelerating insurance workflows, the challenge of maintaining regulatory compliance with automated tools, and the strategic importance of building AI on top of existing legacy infrastructure to unlock trapped data.
Reducing insurance quote times from thirty minutes to under three minutes requires significant automation. How does accelerating this process affect long-term customer retention, and what technical hurdles arise when integrating SMS-based agents with legacy broker management systems?
The leap from a thirty-minute ordeal to a three-minute interaction is more than just a convenience; it is a vital safeguard against customer churn during the most failure-prone parts of the insurance journey. When a process takes thirty minutes, customers often get distracted or frustrated, leading them to abandon the application entirely. By condensing this into a swift, SMS-based experience, we keep the momentum high and the intent focused, which translates into much higher conversion and long-term loyalty. Technically, the challenge lies in the fact that most broker management systems were never designed for real-time, two-way communication. We synchronize these systems by building a reasoning layer that sits on top of them, allowing our agents to pull real system data to answer questions and then instantly push updates back into the CRM or quoting platform as the conversation progresses.
Proactive AI agents now handle routine follow-ups and document collection throughout the policy lifecycle. Beyond simple data entry, how do these tools maintain conversation context across long message threads, and what metrics should firms track to ensure automated interactions don’t sacrifice the quality of customer support?
Our AI agent, Cell, is designed to ensure that conversations stay within a single thread across SMS, iMessage, or RCS, so the customer never feels like they are repeating themselves to a machine. By maintaining this persistent context, the agent knows exactly which documents are missing or which clarification is needed based on previous interactions, allowing the customer to move at their own pace. To measure success, firms should look beyond simple completion rates and focus on engagement metrics, such as how often a customer feels the need to escalate to a human or how quickly a “stalled” conversation is revived by a proactive nudge. We provide insurance leaders with visibility into customer sentiment and specific friction points, ensuring that speed never comes at the cost of the customer feeling genuinely supported and understood.
Insurance distribution relies heavily on licensed professionals and strict regulatory frameworks like RIBO or OTL. How can AI agents be trained to mirror the communication style of licensed advisors while staying compliant, and what are the primary operational risks of deploying these tools in highly regulated jurisdictions?
Training an agent for insurance requires a deep immersion into the specific language of exams like RIBO and OTL, ensuring the AI understands the nuances of coverage as a licensed professional would. We focus on specializing our agents around the way these professionals are taught to communicate, which minimizes the risk of providing inaccurate or non-compliant advice. The primary operational risk in highly regulated jurisdictions is the “black box” problem, where it’s unclear why an AI made a certain statement; we mitigate this by ensuring every interaction is grounded in the carrier’s specific policy data and regulatory guidelines. By making the follow-through automatic and compliant, we allow human brokers to step away from the manual chasing of documents and instead focus on high-level advisory roles where their expertise is most needed.
Many carriers struggle with rigid legacy infrastructure that traps valuable data. Instead of replacing these systems, what are the practical advantages of building a reasoning layer on top of them, and how does this approach change the way large enterprises manage internal data flows and customer intent?
Replacing legacy infrastructure is an incredibly expensive and risky multi-year endeavor that most Fortune 500 companies want to avoid, so the practical advantage of a reasoning layer is immediate utility without the “rip and replace” headache. This layer acts as a bridge, allowing modern AI to “talk” to old data and extract the trapped value that has been sitting dormant in rigid systems of record for decades. It fundamentally changes internal data flows by turning static records into active participants in a conversation, where customer intent is captured in real-time and mapped directly to system updates. This approach allows large enterprises to become AI-native almost overnight, using their existing historical data to power sophisticated, agentic workflows that feel modern to the end user.
Retention rates often suffer when customers shop aggressively during the renewal phase or after a claim. How does automating the coordination of claims and billing workflows prevent revenue loss, and can you share an example of how data from customer questions helps refine broader insurance product offerings?
Revenue loss often occurs not because the product is bad, but because the coordination during high-stress moments like claims or renewals is fragmented and slow. By automating the follow-up and document collection over text, we prevent conversations from stalling, which keeps the customer inside the ecosystem rather than looking for an exit. For example, when an insurer sees through our data that dozens of customers are asking the same clarifying question about a specific coverage limit, they can proactively adjust their policy language or marketing to address that confusion. This flow of data from the frontline “questions” back into the core product strategy allows carriers to refine their offerings based on real-world friction, ultimately making the insurance experience feel invisible and reliable.
What is your forecast for the insurance industry?
I believe we are entering an era where “follow-through” will become entirely automatic and invisible, removing the manual coordination that currently accounts for the industry’s highest operational costs. We will see a shift where the successful carriers and brokers are those who don’t just sell a policy, but maintain a continuous, proactive digital thread with the customer through every stage of the lifecycle. As outsiders and AI experts continue to bridge the gap between legacy data and new intelligence, the “game changer” will be the total elimination of fragmented handoffs, resulting in an industry that finally delivers on the high expectations of the modern, mobile-first consumer.