
The Butterfly in the Machine
August 14, 2026
Why an AI system gives different answers to the same question, what a rounding error in 1961 has to do with your build in 2026, and what it takes to run systems that keep moving.
Industries · Marine Insurance Broking
Under SM&CR a named senior manager carries the accountability for the AI-driven outcomes on your desk today, and the questions underwriters ask insureds about AI were written in London by the market's own committee. What that reaches on the desk, and the safest way to start.
The renewal file keeps taking longer. Regulatory expectations shifted through 2025, the market is remarketing itself as capacity returns, and the same three people carry a book that grew while the profession aged. Brokers under thirty are the smallest share on record.
The record so far
tracked across 25 jurisdictions
Every founding document reviewed by the legally trained founder before delivery.
Documents usually within two business days.
Training certificates independently verifiable at southernsky.ai/verify.
Start Here
Read your brokerage's position in about five minutes. Twenty plain questions in, four things back: the risks as they reach your desk, the regulations that already apply to you, the cost of leaving it unmanaged, and the moves that matter most, ranked.
Get your baselineWhere this sits for you
Meanwhile the tools arrived quietly: a pre-renewal summary here, a policy comparison there, a junior pasting client details into ChatGPT because it was faster. The pattern is documented across markets, adoption running ahead of governance, and the licence obligations were yours throughout. There is a version of this where the admin lifts, the craft stays, and your file notes show who checked what before it reached the client. It starts with knowing your position.
The FCA, the UK's financial conduct regulator, confirmed in April 2025 that it will write no AI-specific rules, supervising AI through Consumer Duty and the Senior Managers and Certification Regime, which it says give it enough regulatory bite, with AI Live Testing running under its AI Lab from autumn 2025. The accountability for AI-driven outcomes already sits with a named senior manager, under rules that already exist.
The LMA, the trade association of the Lloyd's underwriting market, surveyed firms holding over 60 percent of Lloyd's stamp capacity in April 2026: AI moved from limited experimentation to widespread early-stage deployment in twelve months, 93 percent of responding firms run or are building formal AI governance frameworks, and over 60 percent mandate human review of AI outputs. The LMA shipped an AI Adoption Toolkit the same month, governance and accountability first among its five themes.
The LMA's professional indemnity committee, with underwriters from QBE, Dale, CNA Hardy and Aspen, published what to ask insureds about AI, defined use cases, staff training, disclosure of AI use to third parties, oversight checkpoints, and counted 268 legal cases involving GenAI hallucinations since June 2023. The underwriters reviewing your PI renewal have read that report.
BIBA, the British Insurance Brokers' Association, published a members' AI guide in June 2025 and committed in its 2026 manifesto to an AI training school with Markel, while the broker press reports desks feeling the AI questions come thick and fast. A trade body building a school is the demand made visible.
Amiga Specialty secured US$25 million (roughly £19 million) of A-rated Lloyd's capacity for its yachts division within a year of launch, Chaucer and Ceto launched a data-led marine coverholder MGA in March 2026, four new marine entrants joined London in 2025, and the warning in the trade press is that AI adoption is outpacing governance across the MGA market specifically.
Claims turnaround belongs to the insurer, and the daily inbox belongs to you: chasing assessors, parts and repairer shortages, surveyor availability on marine losses. In marine books the client relationship is generational, so a botched claim costs more than the file it sits in.
Documented Work

3 role-based workshop days, designed from a 12-department audit · 18 people certified, from maintenance and marine to finance and the executive team · Every participant left with a working AI setup and a reusable Skill built on a real task from their own role · Weeks later, reported publicly by the client: the team using AI more, sharing wins, and a real shift in the day to day
A twelve-department audit produced three role-based workshop days that certified eighteen people across ten departments, with the CHART quality method and the data rule embedded across the team and every certificate carrying a public verification page.
Read full engagement
One-day executive foundation session, delivered one-on-one, designed around a tightly controlled corporate IT environment where every tool runs in the browser · A monthly management accounting and insurer reporting workflow, sixteen reports each month, run end to end through the CHART quality method and captured as a reusable skill · A take-home toolkit and a short explainer video, so everything built on the day stays usable after it
A one-day, one-on-one foundation session with a senior executive in the marine insurance sector: a governed AI platform with a commercial data agreement behind it, a working environment built from the executive's own context, and a monthly reporting workflow of sixteen insurer reports run end to end and captured as a reusable skill.
Read full engagementThe Opportunity
On the UK desk AI arrived renewal-first: summarising a year of client activity before the renewal conversation, comparing wordings on a remarketed placement, drafting claims chase correspondence, speeding data entry, with the broker checking each output before it reaches the client or the market. The market being placed into has moved further still: most of the Lloyd's market now runs or is building formal AI governance frameworks, human review of AI output is a mandated control at a majority of the firms surveyed, and the underwriters' own committee has published the questions professional-lines insureds should expect.
The firms asking have already built the artefacts they are asking for, and the accountability on your own side is personal: a named senior manager under the Senior Managers and Certification Regime (SM&CR), with Consumer Duty reading outcomes across the file. A written AI policy, an approved-tools register and training records give that senior manager the evidence the regime already assumes, answer the questions arriving on the firm's own PI and cyber renewals, and meet the diligence flowing down from the managing agents in one set. The placement outcome stays the underwriter's call.
Pre-renewal reports
COVA prepares the pre-renewal summary that historically took upwards of 30 minutes per file to understand a year of activity, cutting it to about 15, and the broker checks the summary before it reaches the client.
Policy and wordings comparison
COVA compares policy wordings across markets on a remarketed file, and the broker verifies the comparison before advising.
Market scans
COVA scans the market for alternative capacity when an incumbent moves, and the broker confirms appetite before presenting options.
Client-activity summaries
JAVLN summarises twelve months of client activity for renewal prep and halves the time, and the broker reads it against the file before the renewal conversation.
Claims correspondence
A chatbot drafts claims chase letters and correspondence, and the broker checks the agreed value, the lay-up warranty and the named-storm terms before it goes to the insurer or the client.
Data entry
Juniors use AI to speed data entry, under a rule naming what client detail may go into which tool.
These are the workflows the prompt library and training stand up, under the standard the AI use policy, approved-tools register and training records set, the same artefacts regulators and underwriters ask to see.
Where to start
See what AI could do for your organisation. The AI Baseline Report takes twenty questions and about five minutes, and reads back the use cases returning hours in organisations like yours, the rules that already apply to you, and the moves that matter most, ranked from the top.
The working start for the brokerage: the AI policy, approved-tools register and training pathway a licensee can put in front of the regulator, a PI insurer, or a network audit, with a prompt library that starts the first governed workflow on the renewal desk, 90 days of team education, a recorded briefing, a 30-minute walkthrough call and 30 days of email support. USD $690 founding, then USD $990.
The whole desk to one standard, certificates verifiable at southernsky.ai/verify.
Your marine clients face the same questions; there is a pack for that.
Reading

2026
Federal and state instruments reaching US marine operators.

2026
The regulators, the obligations, and the moves that matter first.

August 14, 2026
Why an AI system gives different answers to the same question, what a rounding error in 1961 has to do with your build in 2026, and what it takes to run systems that keep moving.

Written from inside your world
Kristina Agustin
Founder & Principal Digital Navigator, Southern Sky AI
20+ years in international superyacht and maritime operations. Legally trained (LLB, Graduate Diploma of Legal Practice). AI educator and consultant. ATSE Elevate Scholar 2026.
Questions
The AI Baseline Report reads your position in about five minutes, and your answers pre-fill everything that follows.
Already know where you stand? Governance Essentials puts your position in writing inside 30 days.