The Intake Bottleneck: How MGA and Carrier Pain Points Converge in the U.S. and London Markets

I prepared a study on the submission intake bottleneck facing MGAs and carriers in the U.S. and London markets — where the pain points are, why current automation isn’t solving them, and what the shift toward continuous-learning platforms means for the industry as half its workforce heads into retirement.

Executive Overview

For those of you who can’t read all 8 pages, here are the key points:

The Market Reality:
The U.S. E&S market crossed $105 billion in 2025. MGA premiums hit $114 billion. Lloyd’s wrote £57.9 billion. Submission volumes are at historic highs — and 50% of the insurance workforce will retire in the next 15 years, taking decades of tacit underwriting judgment with them.

The Productivity Gap:
Underwriters spend 70% of their time on non-underwriting work. A single commercial submission contains 300–500 data points; legacy processes capture roughly 50. Manual intake runs 60–140 minutes per submission. In competitive E&S lines, submissions that take 48 hours to process lose to competitors who quote in 24.

Why First-Generation Automation Isn’t Enough:
Traditional OCR and RPA tools extract data the same way every time. They break when formats change. They don’t learn from corrections. They don’t retain institutional knowledge. And they often can’t provide the audit trail that regulators now require — with 25+ U.S. states adopting the NAIC AI Model Bulletin and Lloyd’s mandating CDR compliance through Blueprint Two.

What the Market Needs:
The gap isn’t extraction speed — it’s organizational learning. The next generation of intake platforms must adapt to broker-specific formatting over time, capture the corrections and preferences of experienced underwriters before they retire, and provide full evidence traceability back to source documents. Systems that compound in accuracy with use, rather than degrading when conditions change, represent the structural shift the industry requires.

The Distinct Pain Points by Persona:

• U.S. MGAs — Need to scale GWP without proportional headcount growth while proving underwriting discipline to capacity partners through clean, structured data.
• U.S. Carriers — Need to reclaim underwriter time from administrative tasks and satisfy increasingly rigorous state-level AI governance and audit requirements.
• Lloyd’s Syndicates — Need to map unstructured specialty submissions (slips, MRCs, manuscript wordings) to CDR-compatible structured data without adding operational overhead.
• Managing Agents — Need to reduce the burden of coverholder oversight across highly variable bordereaux and reporting formats.

The Bottom Line:
The insurance industry doesn’t have a data extraction problem. It has a knowledge retention problem. The organizations that solve for compounding intelligence — not just faster processing — will be the ones positioned to scale through the retirement wave.

The Full Article

The commercial insurance industry is facing a structural crisis that cannot be solved by simply hiring more people. Across both the U.S. excess and surplus (E&S) lines and the specialty Lloyd’s of London market, managing general agents (MGAs) and carriers are caught between unprecedented premium growth and a severe talent shortage. As submission volumes surge, the manual intake process has become the primary bottleneck choking profitability, speed-to-quote, and underwriting discipline.

While traditional automation tools have attempted to address this bottleneck, they often fail when confronted with the complex, unstandardized realities of commercial submissions. This article examines the distinct operational pain points facing MGAs and carriers in the U.S. and London markets, the limitations of current extraction tools, and why the industry is shifting toward continuous learning platforms like Cazimir to capture institutional knowledge before it disappears.

The U.S. Market: Growth Collides with Capacity

The U.S. E&S market has experienced historic expansion. Over the past decade, E&S premium volume grew by 223%, expanding from $40 billion to roughly $130 billion in direct premiums written [1]. In 2025, the U.S. E&S market surpassed $100 billion for the first time, reaching $105.31 billion [2].

Simultaneously, MGAs have become a dominant force in distribution. U.S. MGA premiums climbed 16% to $114 billion in 2024, meaning MGAs now account for approximately 11% of the entire U.S. property and casualty market [3]. However, this growth has exposed severe operational vulnerabilities.

MGA Pain Points: Speed, Scale, and Margins

For MGAs, speed is the ultimate competitive advantage. In competitive E&S lines, submissions that take 48 hours to process frequently lose to competitors who quote in 24 hours [4]. Yet, MGAs are expected to scale rapidly without the operational infrastructure of enterprise carriers.

The primary challenges for MGAs include:

• Scaling Without Headcount: MGAs operate on thin margins and cannot afford linear headcount growth to match premium growth. Manual data entry severely limits the number of submissions an underwriter can process.
• Proving Discipline to Capacity Partners: Carriers and reinsurers demand real-time transparency into performance, clean data, and evidence of disciplined risk selection.
• Managing Broker Variance: Retail agents submit data through the path of least resistance. Submissions arrive as unstructured PDFs, mixed emails, and incomplete forms. When MGAs take too long to triage and respond, producers default to faster competitors.

Carrier Pain Points: Underwriter Utilization and Regulation

U.S. carriers face a different manifestation of the same problem. Rather than pure speed, carriers struggle with underwriter utilization and regulatory compliance.

According to longitudinal research by Accenture and The Institutes, the average underwriter spends 70% of their time on non-underwriting activities, including 40% on administrative tasks and manual data entry [5]. A separate study by hyperexponential found that underwriters spend an average of three hours per day strictly on manual data entry [6]. A typical commercial submission contains 300 to 500 pieces of information, yet legacy processes often capture only about 50 of those fields, leaving the rest as unused “dark data” [7].

Furthermore, U.S. carriers are facing an increasingly strict regulatory environment regarding artificial intelligence and underwriting transparency. As of early 2026, more than 25 states have adopted the NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers [8]. This bulletin requires documented governance, risk management, and the ability to explain how AI systems function. “Black box” extraction tools that cannot trace a data point back to its source document expose carriers to significant compliance risks during state Department of Insurance market conduct examinations.

The London Market: Specialty Complexity and Blueprint Two

The Lloyd’s of London market operates differently than the U.S. admitted or E&S markets, but it faces an equally daunting set of intake challenges. In 2025, the Lloyd’s market delivered strong results, with gross written premium rising 4.2% to £57.9 billion [9]. However, the market’s reliance on complex, multi-party placements creates unique friction.

Subscription Market Complexity

Lloyd’s operates as a subscription market, where multiple syndicates each take a percentage of an overall risk. A Lloyd’s broker prepares a “slip” (often aligned to the Market Reform Contract structure), a lead underwriter sets the terms, and “follow” markets subscribe to their share [10].

The submissions feeding this process are notoriously complex. Specialty risks like marine, aviation, political risk, and cyber insurance rely on manuscript wordings, multi-layered slip structures, and varied schedules of value. The lack of standardization makes traditional data extraction incredibly difficult.

Additionally, delegated authority business—where coverholders underwrite on behalf of syndicates—accounts for approximately 45% of Lloyd’s premium income [11]. Managing agents face an immense reporting burden overseeing these coverholders, and bordereaux data quality continues to be a persistent headache across the London market.

Blueprint Two and the Core Data Record

To address these inefficiencies, Lloyd’s launched Blueprint Two, a digital overhaul aimed at replacing manual, paper-heavy workflows with automated, standardized processes. The initiative centers on the Core Data Record (CDR), which mandates the capture of 37 mandatory and 180 conditional mandatory data fields at the point of binding [12].

However, Blueprint Two has faced multiple delays, with full implementation now planned for mid-2025. Market participants are experiencing “change fatigue” and the financial burden of “double paying”—maintaining legacy systems while investing heavily to prepare for CDR compliance [13]. Syndicates need intelligent extraction tools that can map unstructured specialty submissions directly to CDR-compatible structured data without adding operational overhead.

The Universal Threat: The “Silver Tsunami”

Underpinning the intake crisis in both the U.S. and London is a demographic time bomb. The insurance industry is facing a historic talent shortage, often referred to as the “Silver Tsunami.”

• Over the next 15 years, 50% of the current insurance workforce is projected to retire, leaving more than 400,000 open positions [14].
• In the London Market, the situation is particularly acute. More than 25% of underwriters in the City are over 50, and the London Market Group reports that the over-50s and under-30s now each account for roughly the same share (24%) of the talent pool [15].
• A staggering 93% of insurance CxOs are concerned about this knowledge hemorrhage, yet 83% still rely on manual methods (like people-to-people transfer) to capture knowledge before employees leave [16].

When senior underwriters retire, they take decades of institutional knowledge with them—how to read a specific broker’s formatting quirks, which statement of value columns actually matter, and what a “good” submission looks like versus a bad one. Hiring entry-level staff cannot immediately replace this deep, tacit expertise.

The Failure of Static Automation

To solve the intake bottleneck, the industry has heavily invested in Intelligent Document Processing (IDP) and Robotic Process Automation (RPA). By 2025, early adopters of AI-assisted intake reported cutting time-to-quote by 30% to 40% on standard risks [17].

However, traditional automation tools have a fundamental flaw: they are static.

• Brittleness: Legacy OCR and RPA tools extract data the same way every time. They rely on fixed templates and break the moment a broker changes a form or submits a rotated, poor-quality scan.
• No Memory: When a human underwriter corrects an extraction error in a traditional system, the system does not learn. The underwriter will have to make the exact same correction on the next submission from that broker.
• Lack of Traceability: Many generic AI tools operate as black boxes, providing extracted values without linking them back to the source document, violating the explainability requirements demanded by NAIC regulations and internal audit teams.

The Cazimir Solution: Compounding Intelligence

The convergence of surging submission volumes, retiring expertise, and strict regulatory requirements has created the need for a new category of software: the Learning Insurance Operations Platform. This is the exact pain point that Cazimir was built to solve.

Rather than merely extracting data, Cazimir is designed to capture institutional underwriting knowledge. It addresses the specific needs of MGAs and carriers through several core mechanisms:

1. The Learning Loop (Compounding Intelligence)

Unlike static tools, Cazimir learns from every human interaction. When an underwriting team validates results, corrects an error, or resolves a flagged inconsistency, that correction acts as a training signal. Broker-specific formatting conventions and field preferences are learned once and applied forever. As a result, the system’s intelligence compounds; the 100th submission from a specific broker processes significantly faster and more accurately than the first. This directly combats the “Silver Tsunami” by capturing the tacit knowledge of senior underwriters inside the platform before they retire.

2. Handling “Ugly” Submissions

Commercial insurance submissions are rarely clean. They arrive as mixed packages containing ACORD forms, loss runs, and highly variable statements of value, often featuring poor-quality scans or handwritten notes. Cazimir is engineered to classify and extract data from any format, any broker, and any level of quality, adapting automatically without requiring IT teams to reprogram templates.

3. Evidence-First Traceability

To meet the rigorous audit requirements of U.S. state DOIs and Lloyd’s managing agents, Cazimir operates with zero black boxes. Every extracted field links directly to its source document and specific page location. When regulators demand to know how a decision was made, or an underwriter questions a data point, the evidence is immediately accessible. This full auditability ensures compliance with the NAIC Model Bulletin and provides MGAs with the transparent data needed to prove risk selection quality to capacity partners.

Conclusion

The manual intake of commercial insurance submissions is no longer sustainable. As the U.S. E&S market scales and the London Market pushes toward Blueprint Two digitization, the reliance on human data entry is costing carriers and MGAs millions in lost premium and wasted underwriter capacity.

The solution is not simply to automate extraction, but to implement systems that learn. By transitioning from static automation to compounding intelligence platforms like Cazimir, insurance organizations can accelerate quote turnaround, satisfy regulatory audit requirements, and permanently capture the institutional memory of their retiring workforce. In the modern commercial market, the organizations that learn the fastest will be the ones that win.

---

References

[1] MarshBerry. “How The E&S Market Became The Backbone For Emerging Risks.”
[2] S&P Global Market Intelligence. “US E&S industry growth rate dips into single digits in 2025.” (April 2026).
[3] Conning. “2025 MGA Study: Managing General Agents Built for What’s Next.” (July 2025).
[4] Regure. “MGA Claims Platform — Delegated Authority.”
[5] Accenture. “Why underwriters don’t underwrite much.” (February 2022).
[6] hyperexponential. “Why underwriting inefficiency is holding back performance and profitability.” (March 2025).
[7] Accenture. “Intelligent ingestion: The start of commercial digital insurance.”
[8] WaterStreet Company. “What the NAIC Model Bulletin Means for Insurance AI.” (April 2026).
[9] Lloyd’s. “Lloyd’s market delivers strong full year performance.” (March 2026).
[10] Costero Brokers. “How Risk Placement Works at Lloyds of London.” (January 2026).
[11] Lloyd’s. “Delegated Authority at Lloyd’s.”
[12] FinTech Global. “How Lloyd’s Core Data Record is transforming the insurance industry.” (February 2025).
[13] Insurance Edge. “How to Break the Blueprint Two Fatigue.” (April 2025).
[14] U.S. Chamber of Commerce / NAMIC. “The America Works Report: Industry Perspectives.” (June 2021).
[15] RSM UK. “The future of insurance: an ageing workforce and a growing talent gap.” (January 2026).
[16] Insurance Thought Leadership. “Insurance’s Silver Tsunami Knowledge Crisis.” (December 2025).
[17] Pibit.ai. “Submission intake automation in commercial P&C: a 2026 field guide.” (June 2026).