The market that wants founders

Capacity providers say they are cautious about new agencies. Their money says the opposite. The gap between the two is the cost of trust — and that cost is falling.

The previous paper priced the leap from the founder’s side. This one asks why the other side of the table should want the leap made. The answer is in the record, not in sentiment.

01 — The record

Australian underwriting agencies write roughly $10.5–12.9 billion of gross premium a year — about a fifth of general insurance — across some 300 firms. UAC members alone have grown from $7.5 billion in 2022 to more than $10.5 billion today. Sixty per cent of brokers now place more than 30 per cent of their business through agencies. Macquarie Equity Research, which covers the listed insurers rather than the agencies, put it this way in June 2026: agencies are “succeeding in mass-market areas where scale should have made it impossible for them to compete.”

Most of those 300 firms started the same way — an underwriter, an appetite, and a binder. No accelerator or subsidy was involved. One of the most productive founder pipelines in Australian financial services ran itself, unnamed, because specialisation pays.

Offshore it is the same. US MGA premium reached US$108.7 billion in 2025, up 17.8 per cent, a fifth straight year of double-digit growth against 5 per cent for the market. Lloyd’s — £57.9 billion of premium, 87.6 per cent combined ratio, rated A+ — writes around 40 per cent of its business under delegated authority. That is dependence on agencies, not tolerance of them.

02 — What capacity gets

Three things it cannot produce internally.

Appetite. A large book converges on the same risk selection, the same wordings, the same declines. Each specialist who leaves a corporate seat adds a place where some class of risk is priced on its merits.

Distribution. Brokers ring underwriters by name, not logos. A founder answers the phone; the eleventh person in a corporate queue does not. That is why brokers place more through agencies, and why AUB valued Pacific Indemnity, a Melbourne professional-indemnity agency built by underwriters, at A$192 million of enterprise value in 2024.

Cost. The agency carries no prudential capital, no Financial Accountability Regime, and no institutional overhead. Incumbents are shedding that overhead the hard way: Allianz Partners cutting 1,500–1,800 roles, Acrisure 2,250, and 43 per cent of insurers holding headcount flat — a fifteen-year high — with automation the leading stated cause.

03 — What capacity fears

The caution is not prejudice. It is pricing. An unproven agency is expensive to supervise: the founder’s track record sits in a former employer’s systems, the early close is manual, bordereaux arrive late, and one binder is one counterparty. Lloyd’s keeps “a laser focus” on delegated authority and will restrict the right to delegate where coverholders are badly managed; the LMA’s Audit Scope V3.1 now reaches into IT, cyber and AI risk. AFCA expects a “strong, and effectively deterministic” counterfactual when an insurer relies on its guidelines — a system output pointing to decline is not enough.

Supervising a small firm has cost nearly as much as supervising a large one, on a fraction of the premium. On that arithmetic the gate was rational: more founders or more discipline, pick one.

04 — What changed

The gate assumed that evidence of good conduct is assembled afterwards, by hand, at a cost only scale can absorb. That assumption is expiring.

The machinery an agency now rents by the policy — quote, bind, issue, bordereaux, reconciliation — can be built so that the evidence is a by-product of the work: every quote and referral recorded as it happens, rules and wordings versioned, corrections appended rather than overwritten, a close measured in hours. Decisions made by written rules, with the language model confined to reading and drafting, produce AFCA’s deterministic counterfactual on demand. These are design claims about how such infrastructure should work, not a description of any live system.

When the record proves itself, the cost of trusting a three-person firm stops scaling with its headcount and starts scaling with the quality of its machinery — and the machinery is rented, so the smallest firm can hold the same standard as the largest. The machinery that made incumbents big is now rented by the policy, and the discipline that made them auditable is rented with it.

05 — The conclusion

Encouraging entrepreneurship does not mean lowering the bar; a market that admits founders it cannot supervise is arranging its next remediation. It means lowering two tolls: the cost of starting, and the cost of being trusted. Lower the first alone and you get more applicants at the same heavy gate. Lower both and you get more founders and tighter discipline at once, because the discipline is built into what the founder rents. Properly built, the standard is not the obstacle to entry. It is the door in.

Sources & basis. Macquarie Equity Research, Australian Insurance — AI, 23 Jun 2026 (quoted with attribution); UAC (uac.org.au; Insurance News, Dec 2022); Insurance Business AU, May 2026 (broker placement); AM Best MGA Market Segment Report, Jun 2026; Lloyd’s FY2025 results, Mar 2026; AM Best rating Aug 2024; Rachel Turk, Lloyd’s, May 2025; LMA Audit Scope V3.1, Oct 2025; Insurance News, Jun 2024 (AUB / Pacific Indemnity, EV A$192m, 70% acquired); Allianz Partners Jul 2026, Acrisure May 2026, Aon/Jacobson Q1 2026. Statements about self-proving records, versioned rules and close times are design claims, not descriptions of a live system or any named party’s capability. Companion papers: 08 (the end of small), 11 (the declinature drawer), 12 (the pen is portable).

Cuttleflow Systems · Perspective 13 · 33°53′S · 151°16′E · Sydney