This piece is by Drew Young, Founder, Veriscopic
Insurance spends enormous amounts of time and money making decisions.
Underwriters assess risks. Claims professionals interpret coverage and evidence. Delegated-authority teams operate within defined permissions. Increasingly, software and AI systems assist all three.
But there is another cost attached to insurance decision-making that receives far less attention.
It appears after the decision has already been made.
A claim is challenged.
An underwriting decision is reviewed.
A complaint arrives.
An auditor wants evidence.
A reinsurer asks questions.
Litigation begins.
Suddenly the organisation isn’t making a decision.
It is reconstructing one.
At Veriscopic, we call the resulting cost Reconstruction Burden.
And we believe it is becoming an increasingly important operational problem for insurance.
Knowing the outcome isn’t the same as understanding the decision
Most insurance organisations can establish the eventual outcome of a consequential decision.
The claim was accepted.
The risk was declined.
An exception was approved.
A reserve was changed.
Authority was exercised.
The harder question is often:
Why was that decision reasonable given what was known at the time?
Answering that can require considerably more.
What information was available?
Which policy, guideline or rule applied?
What authority did the decision-maker possess?
Which systems contributed?
Were exceptions triggered?
Where did human judgement enter?
Much of this information probably existed when the decision was made.
The problem is that it may no longer exist together, in the state in which the decision occurred.
Instead, evidence is distributed across claims systems, underwriting platforms, emails, documents, model logs, data stores and people’s memories.
The organisation then has to rebuild the decision from fragments.
That is Reconstruction Burden.
Insurance may be paying for the same decision twice
The economics are peculiar.
An insurer invests in systems, data and experienced people to make a good decision efficiently.
Months or years later, it can incur another cost establishing how that same decision was made.
One person searches the file.
Another retrieves information from a system.
Someone finds correspondence.
Operations checks authority.
Compliance becomes involved.
Technology retrieves logs.
Legal may eventually enter the process.
A decision that took minutes can consume collective hours to reconstruct.
Yet that cost rarely appears as a distinct operational metric.
It disappears into claims, underwriting, compliance, audit, operations, technology and management time.
That raises a question Veriscopic believes the industry should start measuring:
How much organisational capacity is being spent reconstructing decisions the organisation has already paid to make?
Upendra Belhe PhD, an adviser to P&C executives on decision ownership in underwriting, claims and AI-driven workflows, captured the economic opportunity neatly:
“Measuring that burden could make decision-state preservation economically visible, not just a governance requirement.”
That distinction matters.
If Reconstruction Burden can be measured, this stops being solely a record-keeping or governance discussion.
It becomes an operational economics question.
AI could amplify the problem
Reconstruction Burden predates artificial intelligence.
AI simply changes the scale and complexity.
Consider an AI-assisted insurance decision.
A model makes a recommendation.
Rules are applied.
A human reviews it.
An exception is identified.
Authority is exercised.
The workflow records an approval.
The audit trail may therefore appear complete.
But an approval trail tells us what happened.
It doesn’t necessarily preserve what made the decision make sense at that moment.
Months or years later, the organisation may need to establish what the reviewer saw, which model or rules were active, what information was available, what exceptions were surfaced and where human judgement influenced the outcome.
Karl Pedersen, CEO of Celadon IP Holdings, whose work focuses on intangible risk, provenance and decision integrity, sees the same challenge emerging:
“The growing use of AI and distributed decision-making makes preservation of context, provenance and integrity increasingly critical.”
This creates a potential paradox.
AI can dramatically reduce the marginal cost of producing decisions while increasing the number, speed and complexity of decisions an organisation may subsequently need to explain.
Decisions could become cheaper to make while becoming more expensive to reconstruct.
From audit trails to replayability
At Veriscopic, we believe the answer is not simply more logging.
Audit trails remain important.
But recording a sequence of events and preserving the state surrounding a consequential decision are different things.
The alternative is to preserve sufficient contemporaneous decision-state so that a later reviewer can understand the basis on which the organisation acted without having to rebuild that state retrospectively.
We describe that capability as replayability.
The distinction is simple:
Reconstruction asks an organisation to rebuild the past.
Replayability preserves enough of the past that extensive rebuilding should no longer be necessary.
This is also why Veriscopic is developing the Verified Evidence Standard (VES): a framework for establishing what evidence surrounding consequential decisions should be preserved so that its context, provenance and integrity can later be independently understood and verified.
The objective is not to preserve everything.
It is to preserve what matters.
Evidence needs to survive organisational boundaries
This becomes particularly important in insurance because the organisation eventually questioning a decision may not be the organisation that originally made it.
Consider:
MGA → carrier → reinsurer
A delegated underwriter makes a judgement.
A carrier assumes the risk.
A reinsurer assumes part of it.
Years later, another organisation may need to understand the basis on which the original decision occurred.
The same problem appears across claims administrators, brokers, outsourced operations and increasingly third-party AI systems.
Insurance therefore doesn’t only have a data-continuity problem.
It has an evidentiary-continuity problem.
Decision-state needs to remain intelligible as responsibility, systems and time move on.
Measure before you transform
This doesn’t require an insurer to begin with an enterprise technology programme.
In fact, Veriscopic believes the better starting point is much smaller.
Take one consequential workflow.
Measure how frequently previous decisions need to be reconstructed.
Count the people involved.
Measure the collective hours consumed.
Identify the systems and evidence sources required.
Separate time spent exercising new judgement from time spent simply rebuilding old context.
Then ask:
How much of that burden could have been avoided if sufficient decision-state had been preserved when the decision occurred?
That creates a baseline.
And once Reconstruction Burden becomes measurable, organisations can decide what actually deserves fixing.
The future of insurance will undoubtedly involve faster decisions, more automation and greater use of AI.
But speed without institutional memory creates another kind of inefficiency.
The organisations best equipped for that future may not simply be those capable of making decisions fastest.
They may be those capable of demonstrating, years later and without expensive reconstruction, what they knew, what governed the decision, who or what exercised judgement, and why the resulting action was reasonable at the time.
That is the shift Veriscopic is working toward:
From reconstruction to replay.
Drew Young is the founder of Veriscopic, which is developing replayable decision-state infrastructure for insurance. Veriscopic is examining Reconstruction Burden across consequential insurance workflows and developing the Verified Evidence Standard (VES) for preserving trustworthy evidence around consequential decisions.
Be the first to comment