Insurance & Underwriting
Underwriting inputs and claims-defensible evidence for the AI liability book
Carriers are writing AI and cyber liability with almost no risk data. Classify how much autonomy a decision actually had, score how far its consequences reach, and evidence the controls behind it, so the exposure can be described in terms an underwriter recognizes.
Challenges We Solve
Common pain points that structured decision models eliminate.
Exposure Nobody Can Describe
Underwriters are asked to price autonomous systems with no equivalent of a credit score or a telematics feed. Classify autonomy level and blast radius per decision so exposure stops being a narrative and starts being a number.
Applications Answered On Faith
Cyber applications ask about MFA coverage, immutable backups, and restore testing. Most answers are asserted, not evidenced. Score the gap between what is claimed and what can be shown before the form is signed.
Claims That Cannot Be Defended
When a loss happens, the evidence chain is usually reconstructed under pressure. Grade integrity chain, provable authority, and documented oversight while there is still time to close the gaps.
Programs Optimized One Line At A Time
Retention, premium, broker fees, and control spend are decided separately and add up to more than they should. Model them as one budget with the loss distribution attached.
Use Cases
How teams use DecisionLedger to make better decisions.
Receives an autonomy classification and blast radius score per decision class instead of a narrative description of the applicant's AI program, and compares them across a book.
Exposure described in terms that support a risk class rather than a judgment call
Runs retention simulation at several candidate self-insured retentions, sizing against the fifth percentile rather than the expected loss, then optimizes the full program budget.
A retention the balance sheet can actually fund in a bad year
Scores evidenced control posture against a typical carrier minimum before the cyber application is answered, and closes the restore-test and privileged-account gaps first.
Application answers backed by artifacts instead of assertions
Measurable Impact
Based on platform benchmarks across early adopters.
Autonomy
Static agent configuration
Classified per decision
Blast Radius
Transient string in a risk check
Scored and persisted
Application Answers
Asserted from memory
Scored against held evidence
Retention Sizing
Expected-loss rule of thumb
Simulated to the 95th percentile
Built for Carriers and Risk Managers
The exposure unit, the control evidence, and the program economics in one governed place.
Exposure Classification
Per-decision autonomy level and blast radius, written onto the decision record so exposure is an attribute of the event rather than a description of the vendor.
Evidence Grading
Control posture and claims defensibility scored against a documented checklist, with the gaps ranked by what a dispute would actually turn on.
Program Economics
Monte Carlo retention sizing, total cost of risk optimization, and captive feasibility on discounted cash flows with the loss distribution attached.
Connects With
Part of 150+ native integrations across CRM, marketing, finance, HR, ecommerce, and analytics
Salesforce
HubSpot
Stripe
Shopify
Google Analytics 4
Workday
QuickBooks
Snowflake
Slack
Zendesk
GitHub
Meta Ads
Mailchimp
NetSuite
Jira
Power BISalesforceHubSpotStripeShopifyGoogle Analytics 4WorkdayQuickBooksSnowflakeSlackZendeskGitHubMeta AdsMailchimpNetSuiteJiraPower BIFeatured Models
Pre-built decision models ready to run with your data.
Autonomy Level Classifier
Assigns a per-decision autonomy level from oversight configuration, override history and gate outcomes, so autonomy becomes an attribute of the decision record rather than static agent configuration.
Blast Radius Estimator
Scores the consequence side of a decision class: reversibility, affected-party count, monetary scope and recovery cost. Blast radius exists today only as a transient string inside risk classification.
Captive Feasibility Analyzer
Compares a single-parent captive, a group captive and continued commercial placement on discounted net benefit over the planning horizon.
Claims Defensibility Scorer
Grades whether the evidence chain behind a loss event would survive a coverage dispute: integrity chain intact, authority proven, oversight documented, timeline reconstructable.
Cyber Insurance Readiness
Maps control posture against the questions on a cyber insurance application and prices the gap between answers given and answers that are actually evidenced.
Deductible & Retention Optimizer
Simulates aggregate retained loss across candidate self-insured retentions to find the retention whose total cost distribution is acceptable at the organization's tolerance.
Errors & Omissions Exposure Model
Scores the professional-liability exposure created by automated advice: volume of advisory outputs, evidence of reliance, disclaimer coverage and correction latency.
Reinsurance Attachment Optimizer
Simulates aggregate loss against candidate attachment points and limits to characterise ceded and retained outcomes for a treaty.
Total Cost of Risk Optimizer
Minimises total cost of risk across retained loss, premium, broker fees and control spend, rather than any single line in isolation.
How It Works
Three steps to structured, auditable decisions.
Classify the Exposure
Assign autonomy level from oversight configuration, override history, and gate outcomes. Score blast radius across reversibility, reach, monetary scope, and recoverability.
Evidence the Controls
Score control posture against the questions an application actually asks, and grade the evidence chain that would have to survive a coverage dispute.
Structure the Program
Simulate retained loss against transfer cost at candidate retentions, then optimize the whole program rather than any single layer.
You are being asked to insure autonomous decisions, or to buy cover for them, and neither side has the data. The controls exist in one system, the decisions in another, and the outcomes nowhere.
Narrative AI questionnaires
Free-text descriptions of governance that cannot be compared across two applicants
Flat-percentage retentions
Retention set by convention rather than by the loss distribution it has to absorb
Post-loss evidence hunts
Reconstructing authority and oversight under dispute pressure, months after the event
Line-by-line program renewal
Optimizing premium and retention separately so the total cost of risk is never actually minimized