General Magic builds insurance-specific AI infrastructure for brokerages and carriers, deploying AI agents across underwriting, service, quoting and claims workflows from Toronto.
General Magic builds insurance-specific AI infrastructure for modern brokerages and direct carriers, deploying AI agents into live workflows across underwriting, service, quoting, claims intake and document collection. The company is based in Toronto and has raised $7.2 million. Its work sits at the hard end of applied AI: not ungrounded chat interfaces, but production systems trained on each brokerage's processes and carrier relationships, connected to email, agency management, CRM, carrier, MGA and broker systems.
The platform functions as an insurance operating system, orchestrating a workforce of specialised agents. Edward supports underwriting decisions with answers grounded in carrier manuals, policy wordings and rate guides. Ava handles routine service requests through voice, text and email. Peter reviews submissions and calls for coverage gaps and errors-and-omissions exposure. Throughout, agents return sourced answers and escalate work requiring licensed personnel to human brokers with full context, preserving the human-AI workflow on which regulated insurance work depends.
The engineering demands are correspondingly specific. The technical stack spans large language models, agent orchestration, APIs, system integrations, evaluation pipelines and browser automation. Building it requires engineers comfortable connecting models to idiosyncratic legacy systems, designing evaluation pipelines rigorous enough for a regulated industry, and working alongside licensed brokers to encode domain judgment into software.
