Jointly AI Launches Autonomous Insurance Brokerage Platform

Jointly AI's platform is constructed around a multi-agent architecture, featuring five specialized AI agents that manage distinct stages of the brokerage process. This system includes an intake agent for initial customer conversations, a research agent to scan the market, a quoting agent that directly contacts providers, an analysis agent powered by a proprietary LLM, and a delivery agent for presenting recommendations. An enterprise-grade orchestration layer coordinates these agents, handling task sequencing, retries, and ensuring auditable, real-time logging of all actions. This move toward autonomous agents reflects a broader trend of deploying multi-agent systems (MAS) in insurance to handle complex, multi-step problems like claims processing and underwriting. Unlike monolithic AI, MAS architecture decomposes workflows into specialized, collaborative agents that can operate independently and in parallel, improving scalability and resilience. Frameworks like LangChain, AutoGPT, and CrewAI are often used to enable the necessary coordination and complex reasoning chains for these systems. For backend engineers, this architecture signals a shift from traditional, siloed systems to a more modular, API-driven approach. Building these platforms requires a microservices architecture where core functions like policy management and claims are independent services communicating via APIs. This design improves scalability and allows for faster development cycles, as individual components can be updated without disrupting the entire platform. An API-first design, treating APIs as strategic products, is critical for both internal innovation and external ecosystem connectivity. From an operational perspective, the platform is designed to address significant inefficiencies, noting that brokers can spend up to 60% of their time on non-advisory, administrative tasks. By automating the process of gathering and comparing quotes, a task that can take hours or days, the system can deliver a full recommendation in approximately 35 to 45 minutes. This level of automation is intended to free up human brokers to focus on complex advisory work and client relationships. The launch comes amid a rebound in insurtech funding, with global investment reaching $5.08 billion in 2025, a 19.5% year-over-year increase. A significant portion of this capital, two-thirds in 2025, has been directed toward AI-focused companies, signaling strong investor confidence in AI-native insurance platforms. This trend continued into early 2026, with US-based AI-driven insurtechs securing multiple mega-rounds of over $100 million. The architecture's reliance on a proprietary large language model, "Jointly Insurance Instruct v1," for its analysis agent highlights the move toward specialized, in-house models. This allows for more nuanced understanding of insurance-specific documents and customer priorities than general-purpose models might provide. However, the "black-box" nature of complex AI systems presents challenges for transparency and auditability, which are critical in a regulated industry like insurance. To mitigate this, Jointly AI's platform assigns a confidence score to each extracted data point and is designed to seek clarification when certainty is low. For a technical founder, this represents a validation of the vertical AI thesis: building AI-native solutions for specific, high-value industries. The key challenge is not just the AI model itself, but the entire orchestration and data processing pipeline that allows agents to interact with legacy systems (like insurer phone lines) and provide reliable, auditable outputs. Success depends on robust data governance, clear KPIs, and often a human-in-the-loop for oversight and to handle exceptions. The platform's API-centric design is crucial for integration into existing brokerage workflows without a complete "rip and replace" approach. This allows for incremental adoption, where the autonomous system can first be slotted into areas with the most significant backlogs, such as lead triage or submission cleanup, before expanding its role. For platform engineers, this underscores the importance of developer experience and designing APIs that are easily consumable by both internal and external systems.