Delegated authority now accounts for approximately 40 per cent of Lloyd's underwritten business, representing more than £22 billion in annual premium — a figure that has more than doubled since 2018 and is projected to exceed 45 per cent of market volume by 2027. The London market's appetite for delegated structures has grown consistently, driven by geographic expansion, niche product specialisation, and the efficiency gains available when underwriting authority is placed close to the risk.

The operational infrastructure supporting that volume has not kept pace.

In the delegated authority space, the bordereaux is still largely a spreadsheet. The submission is still largely a PDF. The audit trail between coverholder and managing agent is still largely an email thread. The policy wording that defines what the coverholder can and cannot do sits in a Word document, often last meaningfully updated years ago. None of these artefacts — spreadsheet, PDF, email, Word file — are readable by an AI system. All of them need to change before AI can do anything genuinely useful in the delegated authority chain.

This is not a new observation. Lloyd's and the London market have been working on this problem for years. The Delegated Data Manager platform went live in 2018. Blueprint Two set out a clear vision for standardised, real-time data flows from coverholders directly to syndicates via API, replacing the manual processing of bordereaux and the rekeying of information that currently consumes significant human effort across the market. DCOM — the Delegated Contract and Oversight Manager — is now mandatory for registering binding authorities. Progress is real.

It is also slower than the external environment is moving.

The scale of the operational problem

To understand what is at stake, it helps to be concrete about the current state of bordereaux processing at a typical Lloyd's MGA or managing agent.

A MGA operating across multiple lines of business might process several hundred bordereaux each month — risk, premium, and claims — from a portfolio of coverholders in different jurisdictions, each with slightly different data formats, different column structures, and different conventions for how they encode the same information. Each bordereaux arrives via email, is manually downloaded, checked, cleaned, and rekeyed into the managing agent's internal systems. A dedicated team handles this work. It is skilled work — bordereaux analysts with deep knowledge of how specific coverholders structure their data, what their error patterns are, and how to reconcile discrepancies. It is also work that exists entirely because the data was not structured correctly at source.

Unitary, an AI automation company working in the Lloyd's market, published data in late 2025 that quantified this problem for one market participant. The firm processed several hundred bordereaux monthly across multiple lines. A team of twelve handled the work. Within five weeks of deploying AI agents capable of reading and processing the bordereaux in any format, 99 per cent of bordereaux were being processed without human involvement. Cost reduction was 30 per cent.

The efficiency gain is significant. The more important observation is what it reveals about where human effort is currently going. Highly skilled bordereaux analysts with deep market knowledge are spending the majority of their time cleaning, reconciling, and rekeying data — work that exists because the underlying data structures do not.

Why this matters beyond operations

There is a governance argument here that extends well beyond operational efficiency.

Delegated authority is a structure built on trust and oversight. A managing agent delegates underwriting authority to a coverholder on the basis that the coverholder will operate within defined parameters — appetite, limits, terms, conditions, exclusions — and will report accurately and promptly on the business written. That oversight depends entirely on data quality. If the managing agent's view of what is in the portfolio is derived from manually processed bordereaux that are weeks old and have been through multiple rekeying steps, the oversight picture is correspondingly degraded.

This is not a hypothetical risk. At ITC London 2026, Andrew Fenton from Hercules and Paul Birmingham from Ecliptic Technology described three structural challenges facing the market: systemic data inaccuracy from continued reliance on manual spreadsheet management; a scale problem where market volumes have outpaced the legacy infrastructure designed to support them; and a governance gap where the manual processing chain introduces processing errors, misapplied terms, and pricing inaccuracies that translate into underwriting leakage. Their assessment: the scale of the market has materially outpaced the legacy infrastructure supporting it.

Oxbow Partners, in their Lloyd's and London Market CEO Agenda for 2025, projected DA business exceeding 45 per cent of market premium by 2027. Their target state for delegated authority is equally clear: coverholders operating as digital front ends for underwriters, with seamless digital data flows from broker to capacity provider. The API Gateway is specifically called out as a required component of the DA technology backbone. The gap between that target state and current reality is the opportunity.

The regulatory pressure is tightening

Lloyd's has been moving in the right direction on this, but the pace of regulatory expectation is accelerating.

Claims has been established as a hurdle principle in Lloyd's oversight of delegated authority — meaning a coverholder's claims performance is now a condition of continued delegation, not just a metric to be monitored. That changes the data quality question from a reporting convenience to a licensing condition. If a managing agent cannot demonstrate adequate oversight of claims performance across its coverholder portfolio in near-real-time, the regulatory exposure is material.

Capacity providers across the market are also raising their data quality requirements independently. The businesses that have scaled their DA books on the back of the hard market are now facing more scrutiny from syndicates and managing agents who want better visibility into what is in their portfolios. That scrutiny requires data that is accurate, timely, and structured — not a spreadsheet that arrives a month after the period end.

The ACORD Core Data Record initiative, which aims to create a standardised data set for insurance transactions, is the structural underpinning of where the market needs to go. When that standard is fully adopted, it becomes possible to connect coverholder systems directly to managing agent and syndicate systems via API, eliminating the bordereaux processing step almost entirely. Lloyd's Blueprint Two was explicit about this ambition.

What coverholders should be doing now

The practical question for a Lloyd's coverholder operating in Australia, New Zealand, or elsewhere in the Asia-Pacific market is how to position for this transition without waiting for the London market to mandate it.

The first step is data quality at source. The bordereaux problem is fundamentally a data capture problem. If a coverholder's policy administration system captures risk data in a clean, structured format from the point of binding — with consistent field names, consistent coding for coverage types, consistent treatment of endorsements and mid-term adjustments — the bordereaux can be generated automatically from that data rather than assembled manually. The work shifts from rekeying to validation.

For coverholders still running on general-purpose tools — spreadsheets, generic CRM systems, policy administration platforms not built for delegated authority — this requires a platform decision. The generation of MGA-specific policy administration systems has accelerated, and several platforms now offer bordereaux generation as a native feature, with Lloyd's DDM integration built in. The investment timeline for these decisions is typically 12 to 18 months. Starting that evaluation now means being ready when the market requires it.

The second step is API readiness. Lloyd's Delegated Data Manager already supports API submission as an alternative to manual bordereaux upload. The DDM/DCOM integration, which went live in 2024, allows binding authority information to flow automatically between contract registration and data reporting systems, reducing rekeying significantly. Coverholders who have connected their systems to these APIs are already operating closer to the target state than those still submitting spreadsheets manually.

The third step — and the one most relevant to the broader AI opportunity — is making coverholder data useful for underwriting oversight. A managing agent with real-time API access to its coverholders' risk data can run continuous underwriting oversight: aggregate monitoring, loss ratio tracking, appetite compliance checking. An AI system operating on structured, timely data can flag emerging issues weeks before they would appear in a manually processed bordereaux. That is a qualitatively different oversight model — and one that changes the value proposition of the coverholder relationship.

The first-mover advantage

The businesses that figure this out first — that make their data structured, accessible, and machine-readable before the market requires it — will have a meaningful head start.

Not just operationally. Strategically.

A coverholder that can demonstrate to a managing agent that it operates with clean data, real-time reporting, and API-native connectivity is a lower-risk counterparty. It requires less oversight resource from the managing agent. It is easier to audit. It creates fewer headaches at year-end. In a market where syndicate capacity is selectively allocated and managing agents are increasingly discriminating about which coverholders they support, that positioning has real commercial value.

The Lloyd's market is heading toward a world where coverholders are digital front ends for underwriting capacity — where the data flows are real-time, structured, and API-native, and where the bordereaux as a concept has been replaced by continuous data synchronisation. That world is not here yet. But the direction is clear, and the infrastructure decisions that determine who arrives there first are being made now.

The spreadsheet has had a good run. Its time is ending. The businesses that recognise that earliest will be the ones writing the interesting DA business in 2030.

Sources: Oxbow Partners Lloyd's CEO Agenda 2025 (Mar 2025), Hercules / Ecliptic Technology ITC London 2026 (Feb 2026), Unitary / MGAA AI Agents in DA (Dec 2025), Lloyd's Delegated Data Manager documentation, Lloyd's Blueprint Two, Davies Insurance Solutions DA analysis (Jun 2025), Pro Global DA Management