Declared against actual
Underwriting works from declarations, and reality diverges immediately. The systems have no place to put what becomes known later, so it goes nowhere, until a claim prices the difference and the renewal repeats it.
Technology for carriers whose real exposure moves between renewals.
What underwriting was told, fixed at binding and true on that day.
What the asset, the owner and the use became in the years afterward.
The distance between the two. It is where the losses live, and it is not in the book.
Risk changes faster than the record.
A policy fixes, at one moment, a description of an asset, an owner, an activity and a price for its risk. From that day forward everything in the environment moves except the record.
The asset changes, the ownership changes, the use changes, the weather changes. The carrier’s systems mostly meet reality again at claim time, which is the most expensive possible reunion. Between underwriting and claim lies the actual business: exposure drifting away from the record, portfolio concentrations forming quietly, evidence of change arriving in fragments nobody joins.
The capability that matters is a representation of the insured world that moves with it: assets, owners, activity and events as dated, evidenced state, so underwriting, portfolio and claims work from what is true rather than what was declared.
None of that replaces underwriting judgment or actuarial method. It changes what they are applied to: a book whose state is current, joined and datable, instead of a stack of declarations aging at different rates.
It also changes what the back book is. Business written years ago stops being an archive to be sampled and becomes a population that can be examined again whenever the carrier learns something new about the risk it took.
Underwriting is a decision about the future, made from a description of the past, and then left alone for a year.
These are not features. They are the entities, relationships and records that carry the environment's state. Kept as a register, each one has a moment where the book fixed it, and something in the world that keeps moving it afterward. The short keys are labels for this page, not names of Fenori products.
Policy is the row this environment fully agrees on, which is why it is set apart above. Every other subject in the register starts drifting away from it the day after binding, and the systems that hold them are rarely built to say by how much.
Six patterns recur across this environment, and they fall on either side of the loss. None of them is solved by buying another point solution beside the core systems.
Underwriting works from declarations, and reality diverges immediately. The systems have no place to put what becomes known later, so it goes nowhere, until a claim prices the difference and the renewal repeats it.
Individual risks are priced well. Accumulations, one region, one supplier, one construction type, emerge across policies written by different teams in different systems, and become visible mainly in hindsight.
Feeds are bought, joined ad hoc and pasted into models. When a figure is challenged, nobody can say which vintage of which source produced it. Enrichment without dating and provenance adds confidence, not knowledge.
The claim file becomes the first place the real asset, use and history are assembled. Prior events, related exposures and in-house knowledge exist somewhere, and arrive after the reserve is set.
A storm, a recall or an insolvency touches many policies at once. Mapping an event to the book is a project each time, run under pressure, when it should be a query the systems already know how to answer.
Disputes turn on what was known and when: at binding, at renewal, at claim. Systems that overwrite state cannot testify, and the institution ends up arguing from correspondence instead of from its record.
A serious carrier already runs substantial technology, much of it load-bearing and expensive to have gotten right. Fenori does not begin by assuming any of it needs to be replaced; the engineering begins with what each system already holds, and what none of them can do alone.
Use what works.
Improve what can work better.
Build what is missing.
Replace only where the outcome requires it.
Almost every system in the column beside this one is a system of record for one moment. The capability that is missing is not another record. It is the layer that keeps those moments joined and dated as they age.
The book should be able to answer before the event does.
Exposure, accumulation and net position as dated, evidenced, computed state: a question the systems answer on demand, not a project the institution runs after landfall.
Five operating areas where the environment’s complexity concentrates. Each opens into the same underlying problem: an insured world described in fragments that have to become one dated, evidenced state.
Declarations enriched into evidenced, dated state, divergence surfaced as it happens, and renewals priced on what the asset became rather than what it was said to be.
Concentrations as computed state across the whole book, continuously, instead of annual studies that are stale before they circulate.
The assembled truth at first notice: asset history, related exposures, prior events and in-house knowledge, present before the reserve is set rather than after.
What was knowable at each moment, reconstructable when it is contested, so the institution argues from its record instead of around it.
Business written through brokers, MGAs and partners reaches the carrier as periodic files. The book’s true state trails its own paperwork, and reconciliation quietly becomes a department.
One policy, drawn against one timeline. The record steps only when somebody writes to it. The world moves continuously, and at notice of claim the difference between the two is what the reserve is actually set on. The figure is a schematic, not a measurement, and not a customer reference.
The systems on the left keep running and nothing is ripped out. The engineered system is the layer across them, and underwriting and reserving judgment stays where it belongs, with the people who answer for it.
None of this is the product. Each capability earns its place in a build when the outcome depends on it, and arrives with the governance the environment expects.
One identity for insureds, assets and counterparties across systems, with provenance kept.
The book as it was at binding, at renewal and at claim, not only as it is today.
Assets, owners, policies and events as connected structure rather than parallel tables.
Footprints joined to the book as computation, not as a project.
Accumulation, gross and net position as deterministic computation.
Calibrated on the carrier’s own history and shipped with their evaluation.
Appetite and authority expressed as executable, versioned rules.
Evidence, reasoning and authority held together in one reconstructable record.
Formed around policy admin and claims, not instead of them.
Fenori’s deepest builds so far were engineered for financial institutions, and they are presented as reference architectures, not customer references. One of them was designed against exactly the problem this page describes, and it belongs at the front of the page rather than in a row of equals.
A carrier’s back book is the part of the business that cannot answer for itself. It was written under earlier appetite, earlier data and earlier judgment, and it is examined by sampling, because examining it properly has always been too expensive. Back-Book Assurance is the reference architecture for the opposite arrangement: when capability improves, the population is re-examined against the improvement, and what the institution now knows is applied to what it already wrote.
The idea reads directly here. A better view of an asset, a corrected ownership chain, a newly understood accumulation: each of them is a reason to look again at policies already on the books, with the record of what was known then still intact. That is also the honest limit of this page. It is a reference architecture, engineered elsewhere, not a deployed product in this industry.
The discipline transfers: a common representation over fragmented records, decisions that remain reconstructable, automation under explicit human authority. What does not transfer is assumed context. The physics here, accumulation, treaties and events that cross the book, have to be engineered specifically, from inside the environment.
Written from the work, not about it: why regulated environments get the technology they get, and what a serious system owes the people who answer for its decisions.
State carries provenance. The record is built to be contested, because eventually it will be.
Enrichment is dated and sourced, never silently overwritten by the next refresh.
Gross and net views are first-class, not spreadsheet afterthoughts assembled per event.
At binding, at renewal, at claim: the system answers as of a date, by design rather than by reconstruction.
Models ship with evaluation, consistency testing and failure conditions, or they do not ship.
Policy administration and claims stay where they are. The representation forms around them.
Describe, in plain terms, a decision, an exposure or a process your current systems do not handle well enough.
What should your carrier be able to do that it cannot do today?
This is the point where we would rather understand it properly than pretend we know the answer from a browser.
What should your book be able to tell you before the event?
Describe an exposure, an accumulation, or a claims pattern your systems only reveal in hindsight. Nothing confidential is needed at this stage.
Begin with a challenge