Could Sam Altman's one-person unicorn prophecy realise itself in the form of an MGA (Managing General Agent)?
Even the most bullish AI advocates may hesitate at the idea of a single person running an insurance company; let alone one valued at $1 billion. But what about a managing general agent (MGA)? Could the lean, flexible MGA model, powered by agentic AI, make that a plausible reality?
This piece examines the limits and possibilities of AI in the MGA model: how far one underwriter could scale alone, how many submissions they could realistically process, and where the risks and breakpoints lie.
Sam Altman popularized the notion that generative and agentic AI might enable the world’s first “one-person unicorn”: a founder-only business worth more than $1 billion. At hyperexponential, we are building AI-driven pricing and underwriting technologies that augment actuaries, underwriters, and insurance leaders - turning productivity gains into measurable profitability. Which raises the question: could an MGA fulfil Altman’s prophecy?
How can MGAs leverage AI today?
Many MGAs already exploit AI to improve speed and profitability. Yet their real value continues to flow from relationships and reputation. Capacity origination, negotiations, broker production, bordereaux, and claims all rely on trust and judgment.
One‑person insurance businesses just isn’t a thing for a reason. It’s a very people‑intensive business
Darren Govender, Lead Pricing Actuary at hyperexponential
AI cannot replace those human foundations, but it can amplify them. For seasoned underwriters, codified guidelines and guardrails allow AI agents to:
Ingest messy submissions and map fields
Cleanse, reconcile, and validate data; chase missing information against binder rules
Run approved models, help actuaries iterate and adjust, and auto-assemble rationale
Surface portfolio-level analytics, covering rate adequacy, accumulations, directly at the point of pricing
Generate documents and write outcomes back into MI and audit systems
If you can ingest submissions quickly, the speed of putting out a quote is significantly higher… you don’t need someone to spend time keying that information in. If you can write more with less, why wouldn't you?
Kamlesh Walia, Senior Pricing Actuary at hyperexponential
The question isn’t whether AI can replace MGAs. It’s how far it can increase submissions-per-underwriter, shorten cycle times, and sharpen outcomes.
Where AI fits in the MGA workflow
On paper, an MGA’s workflow looks linear: intake → triage → pricing → portfolio checks → approvals → quote/bind → bordereaux/audit → claims. In practice, each step contains multiple sub-processes scattered across emails, spreadsheets, raters, and portals. This fragmentation throttles throughput.
Submissions arrive in inconsistent formats, forcing manual reconciliation
Clearance and sanctions checks stretch across emails and spreadsheets
Pricing runs in parallel tools, creating version control risks
Portfolio checks and referrals often land too late
Document assembly and MI reconciliation remain manual and error prone
The result: bottlenecks, missed business, and weak auditability.
The best MGAs do two things: stay lean and write profitably. Deep expertise and a tight workflow beats headcount.
Kamlesh Walia
What an AI-Native MGA looks like
An MGA will always build its moat by concocting a unique blend of data, expert judgement, and exceptionally strong relationships. AI can sharpen an MGAs ability to build out strong capabilities in each area:
Data: capturing more, granular, instantly reconciled and cleansed data helps create a data asset that you can orchestrate and build reports on.
Judgement: guardrails, insights at point of pricing, scenario testing all sharpen decisions.
Relationships: faster, more consistent submission-to-quote times strengthen broker and carrier trust, with audit trails to ease capacity discussions.
It’s all well and good to say you’ve got an AI that ingests with 95% accuracy. The risk is that the 5% you miss could misprice a book catastrophically.
Kamlesh Walia
Leaner, smarter MGAs; not quite 1-person unicorns
A fully agentic, one-person $1 billion MGA is highly unlikely. Insurance remains people-intensive. But the upside is still enormous: underwriters can work through more submissions, generate more quotes, compress their submission-to-quote time, all while enhancing risk selection.
Realising the value to AI requires a pragmatic path:
Prioritize production use cases: submission ingestion, governed pricing, real-time portfolio signals, MI and audit baked in
Build the data foundation: capture once, cleanse, reconcile, and maintain lineage
Encode judgment and guardrails, then let AI amplify the stack
Powered by AI, the smaller, leaner MGAs can start to perform and compete with their much larger counterparts. Embracing AI is not going to be a competitive advantage for long, it could become tablestakes. We see a future where AI complements your competitive edge, helping you innovate faster, capitalise on the inherent agility of an MGA and build inherent value in your underwriting processes.
hyperexponential works closely with MGAs to augment their pricing and underwriting capabilities; improving their judgement, enhancing their relationships, and driving P&L performance. Get to see hyperexponential’s AI-powered platform by booking a demo with us today.
