Scy - Executive summary
AI-powered risk intelligence that closes the gap between what you model and what your policy pays.
THE PROBLEM
Insurance for large physical assets is priced using broad industry categories, not the actual quality of a specific company's risk. Well-managed operators overpay by 30-50%. Coverage gaps of 16-40% are discovered at claims time. The broker confirms the quote as "competitive." Generic AI tools summarise documents but cannot reason about insurance - they start fresh every session.
THE PRODUCT
Scy is an autonomous AI agent that builds and maintains a Digital Risk Twin - a compounding knowledge base connecting every location, policy clause, regulatory constraint, and loss scenario. It diagnoses coverage gaps, recommends remediation with cost-benefit analysis, and drafts the documents to execute. Every document processed makes future analyses smarter.
Compounding knowledge + insurance DNA
Every contract, policy clause, PML scenario, and monitoring alert adds context to all future analyses. Scy accumulates institutional memory. Founded by the team that built Swiss Re's physical risk tooling (incl. JV with Palantir) and operators that saved millions in total cost of risk within real-asset-heavy industries such as chemical manufacturing and mining. A purpose-built risk & insurance reasoning engine.
VALUE DELIVERED
30x
ROI - early adopters report 30x return within the first renewal cycle. Savings dwarf the subscription in the first quarter.
7-figure
Annual savings - for companies with EUR1-10M+ PDBI programme, compounded year over year as the risk profile and knowledge base improve.
< 1 cycle
Payback - first renewal delivers both the premium reduction and the closed coverage gaps that justify the platform.
100%
Auditable - every saving traces to a specific clause, scenario, or mitigation. No black box, no guesswork.
Gaps closed before placement
Sublimit mismatches, definition gaps, and regulatory constraints resolved before going to market - not discovered at claims time. Better data unlocks better terms, not just lower cost.
Submissions underwriters read
Quantified, auditable risk profiles in the metrics underwriters rely on. Removes the ambiguity that drives defensive pricing. The submission becomes a negotiation tool, not a formality.
Savings fund prevention
Scy identifies where physical mitigation yields the most premium reduction. Improved risk profile feeds back into lower premiums - a virtuous cycle replacing the fear tax.
30-50%
Premium reduction - evidence-backed submissions replace generic pricing. Seven-figure annual savings.
Seconds
Per what-if question - retention changes, sublimit scenarios, programme restructuring. Replaces 1-2 week broker round-trips.
90 → 0 days
Renewal scramble - continuous preparation replaces the annual data collection sprint.
The transformation: Moving enterprise risk & insurance from a cost center running a 90-day renewal scramble to a strategic resilience capital allocator with real-time intelligence, year-round.
Scy - Risk intelligence platform
You quantify risk well.
You don't know what your policy actually covers.
Scy quantifies your actual risk, checks what your insurance programme covers against it, shows you how to optimise your structure, and tells you where to invest in resilience - continuously, not once a year at renewal.
Scy builds a Digital Risk Twin of your entire portfolio - every location, every policy clause, every regulatory constraint - queryable in real time. Every document it processes makes future analyses smarter. The knowledge compounds.
The problem
Risk quantification is the easy part.
What happens after is where the money disappears.
You produce a clean risk picture. Then three things destroy it.
The fear tax
Generic pricing punishes good operators
A manufacturer with world-class fire suppression and a pristine loss history pays the same rate bucket as a poorly-maintained peer. Without site-level evidence, underwriters price defensively. The good risk subsidises the bad.
Wording mismatch
Policy language doesn't match your risk model
You assume a loss is covered. The 100-page contract says otherwise - through an exclusion, a sublimit cap, a definition mismatch, or a condition precedent you missed. You discover this when a claim is denied.
Regulatory erosion
What you can buy ≠ what you need
Non-admitted rules, tariff floors, exportability restrictions, government pool mandates, and local tax regimes make it impossible to place the coverage your model prescribes. Global programs fragment on contact with local reality.
"The legalistic language is too complicated for me as a risk person. The product I really miss is: how can I translate a 100-page insurance contract into a one-page coverage map against my 10 modeled loss scenarios?"
The problem
You manage a multi-million insurance programme.
You can't prove what it actually covers.
Enterprise risk teams quantify hazards, negotiate renewals, and report to the board - all without the data to verify whether their policies pay what their models predict. The gap between modeled loss and actual recovery is discovered at claims time, when it costs millions.
82%
Lack quantitative evidence when presenting risk quality to underwriters. They bring narrative and broker-relayed numbers to a table where the other side has actuarial models, loss curves, and pricing analytics.
75%
Wait minimum 1-2+ weeks - or don't ask- for a quantitative answer on programme alternatives. "What if we changed our retention?" is a question many have stopped asking because the turnaround makes it impractical.
80%
Can't report impact to the board in numbers. CFO and group risk get narrative summaries and broker reports - not exposure-adjusted TCOR (Total Cost of Risk), not premium adequacy, not coverage position by jurisdiction.
58%
Cannot quantify engineering ROI.Sprinkler upgrades, maintenance protocols, site hardening - they can't prove the premium reduction these generate. Good mitigation projects die because the financial case doesn't exist.
40%+
Of pre-renewal time burned on admin - data collection, SOV assembly, chasing site responses, reformatting spreadsheets. Not risk analysis. Not negotiation strategy. Data logistics.
88%
No C-suite dashboard for TCOR and exposure. And 88% say they need one. The insurance function operates without the reporting infrastructure every other finance function takes for granted.
What this looks like in practice
Energy infrastructure operator - critical pipeline
What they have: NATO-grade geopolitical data feeds, real-time ship tracking, physical pipeline intelligence, proprietary hazard alerts, dedicated risk engineering.
Global manufacturer - 30+ country portfolio
What they have: Internal technical pricing models, CHF 25M+ annual premium, a captive, a dedicated team, engineering data across every site.
The industry treats this as normal. It is not. The gap between what you think your policy covers and what actually pays out is structural, predictable, and preventable. It has never been economically feasible to close - until now.
Source: ScyAI Insurance Manager Pain Points Survey (n=72), Insurance Team Priorities Survey (n=17), Risk Awareness Week 2025 Conference Survey (n~80). Case studies from ScyAI client engagements.
The cost
Millions disappear - in coverage gaps, leakage, and wasted time.
16-40%
Unrecoverable loss - the gap between modeled loss and what the local policy actually pays. Discovered at claims time. Includes tax exclusions, sublimit caps, definition mismatches.
30-50%
Premium overpayment - when independent quantification replaces generic pricing. Early adopters report seven-figure annual savings by submitting evidence-backed risk profiles instead of broker narratives.
1-2 weeks
To answer a single what-if- "What if we changed our retention?" requires a round-trip to the broker. Too slow for iterative decision-making at renewal.
20-40%
Team time on admin - data collection, SOV preparation, and submission assembly in the 3-4 months before renewal. Time diverted from analysis and negotiation.
CHF 0.5-1M
Administrative leakage per year - on CHF 25M premium volume. Commissions, trapped premium, regulatory compliance, cross-border transfer friction, accounting overhead.
9 mo+
Claims dispute cycle - when a denial triggers a dispute, resolution takes 6-12+ months. Internal stakeholders lose confidence in the insurance function entirely.
How it's solved today
Every current path fails at the same point.
No existing solution connects the risk model to the policy language to the regulatory constraint - simultaneously, across every jurisdiction.
Broker
Wrong incentives
Paid percentage of premium. Confirms pricing as "competitive" when the client is overpaying by multiples. Cannot hold the quantitative model and the legal text simultaneously. No regulatory depth per jurisdiction.
Coverage lawyer
Expensive, slow, narrow
Reads the wording. Doesn't understand the loss model. Cannot run scenarios. Charges by the hour. Economically infeasible to review every clause against every scenario annually.
FLEXA & NatCat models
Commodity, disconnected
RMS, Moody's, Air - they quantify hazard exposure. They don't connect to policy wordings, don't know your sublimits, don't check regulatory constraints, don't simulate claims.
Generic AI
No memory, no insurance reasoning
Copilot, ChatGPT, Claude- they summarise documents but start fresh every session. No accumulated context. They flag terrorism limits as too high based on offshore losses. They miss that broad damage descriptions benefit the insured. Confident wrong answers with no institutional memory.
What changed
AI can now reason across all dimensions at once.
For the first time, a system can read a 100-page policy document, parse its clause-by-clause coverage logic, cross-reference it against loss models and regulatory databases, and return a financial position - in seconds, not weeks.
Before
Sequential, siloed, manual
The actuary builds the model. The lawyer reads the wording. The broker checks the market. The regulatory specialist checks compliance. Nobody holds it all together. Gaps live between the handoffs.
The shift
AI + structured enterprise data
AI can parse legal text, reason about quantitative scenarios, and cross-reference regulatory constraints. Combined with a structured knowledge base of your portfolio, this creates the integration layer that never existed.
Now possible
Real-time, multi-dimensional queries
Ask a natural-language question. The system reasons and acts across your locations, loss models, policy clauses, and jurisdictional rules simultaneously - and returns a financial position, not a risk score.
Scy - System architecture
Quantify. Validate. Optimise. Invest. Monitor.
Scy - The experience
One question. Every dimension answered. Resolution proposed.
You ask in natural language. Scy diagnoses the gap, recommends how to close it, calculates the cost-benefit, and asks permission to start the work. From question to action in one conversation.
Why Scy
Built for the insured. Not for the insurer or the broker.
Most risk analytics tools are built for the sell side. Generic AI tools lack insurance reasoning entirely - they summarise documents but cannot think about coverage. Scy is purpose-built for the company that owns the assets and pays the premiums.
Without Scy
With Scy
Compounding knowledge
One source of truth of all your risk context
Every contract, policy clause, PML scenario, and incident monitoring alert adds context to all future analyses. The knowledge tree expands with use. Generic AI tools start fresh every session. Scy accumulates institutional memory - your institutional memory.
Insurance DNA
Built from inside the industry
The founding team operated inside underwriting and risk modelling - Swiss Re, Palantir joint venture. The insurance reasoning engine is not a wrapper on a generic LLM. It's a purpose-built system that thinks in sublimits, exclusions, and jurisdictional constraints.
Submissions, not dashboards
Actionable output, year-round
Scy produces renewal submissions, underwriter rationales, and coverage position documents autonomously with the right guidance at the right time giving you 10x leverage on your time. Scy continuously scans for risk events, regulatory changes, and emerging threats. You get warnings when something material changes - not a quarterly report.
Value delivered
From cost center to strategic resilience capital allocator.
Every output is measurable. Every saving is auditable. The transformation is from reactive insurance buyer to proactive risk strategist.
30-50%
Premium reduction - by replacing generic pricing with evidence-backed, site-level submissions. Early adopters report seven-figure annual savings within the first renewal cycle.
90 days → 0
Renewal scramble eliminated - Scy prepares renewal submissions autonomously, year-round. The 3-4 month data collection sprint becomes a continuous, always-current process.
1 source
Compounding knowledge base - every document, scenario, and market event makes future analyses smarter. Replaces scattered spreadsheets and broker emails with a single queryable intelligence layer that accumulates institutional memory.
Coverage adequacy
Gaps closed before placement
Better data unlocks better terms - not just lower cost but more coverage. Sublimit mismatches, definition gaps, and regulatory constraints identified and resolved before going to market.
Underwriter trust
Submissions they actually read
Quantified, auditable risk profiles using the same metrics underwriters rely on. Removes the ambiguity that drives defensive pricing. The submission becomes a negotiation tool, not a formality.
Resilience capital allocation
Savings fund prevention - and prevention funds savings
Scy identifies where to invest in physical mitigation for maximum premium reduction: fire suppression, structural upgrades, monitoring systems. The improved risk profile feeds back into lower premiums. A virtuous cycle replaces the fear tax.
Next steps
See your portfolio through Scy.
We build your Digital Risk Twin using your actual data - locations, policies, claims, programme structure. You ask it the questions that currently take weeks. You get answers in seconds.
Step 1
Data intake
You provide your site register, master and local policy wordings, claims history, and programme structure. We ingest, parse, and contextualize everything into your Digital Risk Twin.
Step 2
Pilot workflow
We run a policy audit on one line or one jurisdiction of your choice. You see clause-by-clause coverage mapping, gap quantification, and claims simulation against your actual modeled scenarios.
Step 3
Full deployment
Expand across all lines, all jurisdictions. Scy becomes your standing intelligence layer - queryable in real time, updated as policies change, and ready for every renewal cycle.
The pilot takes 2-3 weeks.Pick the jurisdiction where you have the least confidence in your local policy. That's where Scy shows you the most value the fastest.
Try the world's most advanced risk intelligence.
bernhard@scyai.com - scyai.com