07 — Insurance
Submission to settlement modelled as one value engine: what you take on, what you charge for it, what it costs to settle, and what leaks in between — under conduct and rate-fairness discipline.
North Star
Combined ratio per book of business
The value engine, end to end
The same six moves we run everywhere, expressed in the physics of this industry.
Where does the combined ratio actually leak?
Leakage map by product and channel
What drives loss, expense and lapse?
Causal loss and expense model
What happens to the book if we move appetite or rate?
Portfolio and claims twin
Who triages, prices and settles at volume?
Governed underwriting and claims agents
Can we defend every rate and every denial?
Conduct and rate-fairness control file
Does the loss ratio hold through the cycle?
Outcome contract + cohort monitor
Where the value is
Before a twin is built, we agree where the money sits and what moves it.
Value pool
Underwriting selection
Driver
Risk differentiation and appetite discipline
Measured as
Loss ratio points
Value pool
Claims settlement
Driver
Cycle time, reserve accuracy, leakage
Measured as
Cost per settled claim
Value pool
Expense base
Driver
Straight-through processing across submission and service
Measured as
Expense ratio points
Value pool
Retention
Driver
Renewal experience and rate shock management
Measured as
Retained premium per renewal
Use cases
Each one is a number somebody already owns, a twin that models it, and an agent that moves it inside a defined authority. Filter by function.
8 use cases
Reads the claim, the photos and the call transcript at intake, sets the reserve range, and routes straight-through, to an adjuster, or to fraud — in minutes rather than days.
Flags the claims worth investigating and, just as importantly, clears the honest ones fast so good customers are not punished for a fraud programme.
Predicts which claims will be denied, why, and what evidence would change the outcome — then assembles the appeal pack automatically.
Prices risk on the exposure actually present rather than on the segment average, with rate-fairness checks and regulator-ready reason codes.
Finds the coverage a household or business is missing and the bundle that is genuinely better for them, so cross-sell survives the renewal conversation.
Detects the signals that precede a lapse — service friction, rate shock, life event — and triggers the treatment that actually retains, priced against the customer's value.
Reads the claim file for third-party liability that was never pursued and packages the recovery case with its evidence.
Turns broker submissions — email, spreadsheet, PDF schedule — into structured, ratable risk data, so underwriters spend their day on selection rather than on retyping.
Outcome twins
Each twin owns a set of levers. Each feature moves one lever. Each agent executes inside a defined authority.
Simulate appetite, rate and channel mix against the loss ratio that would result.
Levers
Features
Multi-agent layer
Submission Agent
Turns broker packs into structured, ratable risk data.
Referral Agent
Routes out-of-appetite risk with the rationale attached.
Settle the routine fast, reserve the complex correctly, investigate only what pays.
Levers
Features
Multi-agent layer
Triage Agent
Sets the initial reserve range and routes the claim in minutes.
Recovery Agent
Builds the subrogation case with its evidence.
Find organised and opportunistic abuse without punishing honest claimants.
Levers
Features
Multi-agent layer
Investigation Agent
Pre-investigates and packages referrals for SIU review.
Keep the premium worth keeping, at a defensible price.
Levers
Features
Multi-agent layer
Renewal Agent
Prepares the renewal conversation with the customer's own numbers.
Unit economics
Set your volume, switch features on or off, and move each impact to your own evidence. Everything is expressed per policy.
Policies in force
380,000
Contribution per policy
$60
Per-policy contribution; substitute per-claim economics for a claims-only engagement.
Annual program investment
$3.5m
Underwriting Twin
Volume — Submissions quoted
4.0% — Underwriters spend the day on selection, not retyping.
$912k
annual value
Claims Twin
Unit cost — Cost per settled claim
5.0% — Straight-through settlement for the routine majority.
$16.7m
annual value
Claims Twin
Leakage — Claims leakage
22.0% — Reserve and settle on evidence rather than on habit.
$6.3m
annual value
Underwriting Twin
Unit revenue — Premium per policy
2.5% — Price precision inside filed and fair bands.
$8.9m
annual value
Retention Twin
Volume — Retained premium
2.5% — Intervene before the rate-shock renewal, not after.
$570k
annual value
Value by feature
Annual value contributed by each active feature at current settings.
Value and ROI vs volume
How engineered value and ROI move as policies in force change. The marker is your current setting.
Cumulative value vs investment
Where the curve crosses the investment line is payback.
Value mix by lever
Which lever the engineered value is actually coming from.
Baseline contribution
$22.8m
Engineered annual value
$33.4m
ROI on program
855%
Payback
1.3 months
Directional model. In an engagement every number here is replaced by your measured baseline, signed off by finance, and written into the outcome contract before any agent goes live.
Scale
Champion/challenger on a bounded book with a clean baseline.
Same twins, new exposures; rate-fairness artefacts reused.
Shared agent runtime, per-jurisdiction conduct guardrails, one loss-ratio number.
Integrations
Vendor-agnostic by design. Connectors are added per engagement — no platform lock-in.
Policy & claims
Risk & fraud
Data & cloud
Distribution & service
Regulations and standards
Compliance is a design input, not a review gate. Every agent action is logged, attributable and reversible.
EU AI Act
Life and health risk pricing is high-risk: oversight, logging, accuracy evidence.
Conduct rules (FCA / NAIC)
Fair value, vulnerable customers, defensible claims decisions.
Solvency II / IFRS 17
Reserving, capital and contractual service margin consistency.
GDPR Art. 22
Automated pricing and denial rights, explanation and human review.
Rate fairness / anti-discrimination
Proxy testing on every rating and claims model.
Standards supported
Next industry
Telecom & Media
