The Great British Paper Trail Has Met Its Match

In the London insurance market, vast sums still pass through what resembles a slightly modernised antique ritual. The desks are digital; messages arrive via API rather than courier; and no one has worn a bowler hat unironically in years. Yet the underlying choreography remains Edwardian: one party makes a consequential decision, another assumes it was done properly, and a third—often years later, usually under regulatory duress—asks where the evidence might be.

At which point, everyone adopts the expression of a person unfairly accused of misplacing something they were never formally given.

It would be comforting to blame this on a failure to digitise, but that won't do. Insurance has digitised repeatedly, sometimes with an enthusiasm that leaves entire departments quietly exhausted. The industry now possesses platforms, standards, portals, dashboards, and transformation programmes in such abundance that one suspects the real scarcity lies in weekends. Yet something essential is missing: a portable, reliable record of why a decision was made, who made it, under what authority, and what the conclusion was.

Data, in other words, moves rather well. Decision context does not.

This sounds technical until one remembers what insurance is. It is not the processing of documents, but the organised distribution of trust under uncertainty. A loss adjuster's report travels through insurers, brokers, lawyers, and reinsurers gathering annotations and losing clarity in equal measure. At every stage, the question is not what happened, but whether the person who decided was entitled to do so—and whether that decision can be defended.

For decades, the system worked because the evidence, though rarely recorded, was remembered. Senior practitioners carried the context in their heads: who reviewed what, what caveats were raised, what was agreed in the call that never made it into the file. The market functioned not because it was structurally sound, but because enough experienced people could reconstruct the truth when required. It was, in effect, an oral tradition with quarterly reporting requirements.

The difficulty with oral traditions is that they tend to be forgotten.

Many of the industry's daily frustrations share one origin. Rework, duplicated verification, opaque handoffs, audit friction, and the need to ask each participant to explain their decision years later are not separate inconveniences. They are symptoms of a structural absence: the lack of a decision-evidence layer that can travel across firms, platforms, and jurisdictions.

Nowhere is this more visible than in the subscription market. The lead-and-follower model once possessed a certain elegance: the lead did the work; the followers trusted the lead. It was, in the right lighting, almost civilised. That elegance frays when dozens of insurers attach to a claim, portions of the risk have been transferred elsewhere, and justification is demanded not by a familiar counterparty but by a regulator with a deadline.

The era of the handshake has not ended because people have become less trustworthy. It has ended because trust, on its own, no longer satisfies the evidentiary standard.

The remedy is not another platform, nor a centralised system, nor any of the more theatrical cures in fashion. It is something simpler and more disruptive: a structured, signed, portable record attesting to a decision—who made it, on what basis, and with what authority—without revealing everything else. A modern seal: not a wax impression, but a verifiable signal that a decision exists and can be trusted.

Full transparency is not required. The market has always operated on controlled asymmetry—leads maintain narrative control, brokers manage relationships, reinsurers carry risk without direct visibility, and lawyers, it must be said, do not suffer unduly from ambiguity. Any workable architecture must preserve these asymmetries while letting participants verify that proper work has been done.

The timing is, in a word, inconvenient. Three pressures converge. The first is demographic: the practitioners who carry the system's memory are leaving faster than they can be replaced. The second is regulatory: authorities now expect not just proof that actions were taken, but evidence of intent and reasoning. The third is technological: AI is entering underwriting and claims at a pace that leaves little room for workarounds.

AI sharpens the problem: a human approving a machine-generated recommendation without context has preserved neither judgment nor liability. Without a record of what the machine proposed and why the decision was reached, the system does not produce efficiency—it produces future disputes at scale.

The instinctive response will be incremental—one firm connected to another, one integration at a time. These efforts may succeed locally but recreate fragmentation in more elaborate form. A network of bilateral fixes is not a system; it is a patchwork that eventually resembles the problem it set out to solve.

Faintly comic: an industry devoted to pricing uncertainty has left its own decision-making uncertain. The humour is unlikely to endure. When regulators, machines, and the next generation of practitioners all begin asking the same question—who decided, on what basis, and can you prove it?—The traditional answer will not suffice.

"We've always managed" is not, and never has been, an audit trail.