# General Magic: Insurance work with a broker handoff

Canonical: https://mudpie.ai/companies/general-magic/
Breadcrumb: [Home](https://mudpie.ai/) / [Companies](https://mudpie.ai/companies/) / [General Magic: Insurance work with a broker handoff](https://mudpie.ai/companies/general-magic/)
Author: Ali Abouelatta (https://mudpie.ai/authors/ali-abouelatta/)
Published: 2026-09-20
Updated: 2026-09-20
Research type: Company profile
Method: Public-source research. Official Speedrun and company pages checked September 20, 2026. Product claims are attributed to their sources.

General Magic is building insurance-specific agents for brokerages and carriers.

## What it does

The company describes three roles: Edward, an AI Underwriting Officer; Ava, an AI Customer Officer; and Peter, an AI Compliance Officer. Edward works from carrier manuals, wordings, and rate guides, answers appetite and eligibility questions, and shows the source behind an answer. The customer workflow reaches across email, text, and voice, with a handoff to a licensed broker when binding is needed.

The mechanism is therefore source-linked assistance plus a human handoff, not an autonomous promise to bind every risk. The homepage also says a team can see conversations in real time and intervene when a case is high-stakes, frustrating, complex, or risky. Those are product claims; the page does not establish underwriting accuracy or regulatory approval.

## Buyer and task

The buyer is an insurance brokerage or carrier trying to make manuals and servicing knowledge usable across a team without removing licensed judgment. Start with one carrier’s appetite guide and a small set of cases: clearly eligible, clearly outside appetite, and ambiguous.

The important questions are operational. Can the answer show the passage it relied on? Can an administrator limit which manuals or rate guides an agent can use? Does an ambiguous case reach a licensed broker with the conversation and source context intact? A fast answer without those boundaries can make an incorrect eligibility decision harder to spot.

## The first test

Give Edward the three cases and require a source-linked response for each. Ask it to answer a servicing question over the channel the team actually uses, then deliberately interrupt with a case that requires human judgment. Review the handoff, the retained context, the audit record, and the point at which the agent stops.

The company’s privacy policy says its messaging agents may connect to CRM, policy, and claims systems and says General Magic maintains SOC 2 Type II compliance. Those statements should be verified against the customer agreement and the specific carrier systems before production use.

## Sources checked

[official Speedrun profile](https://speedrun.a16z.com/companies/general-magic) · [company homepage](https://generalmagic.inc/) · [privacy policy](https://generalmagic.inc/privacy-policy)

Sources checked — September 20, 2026.


## Author disclosure

I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.