Watch your war-risk book on live data —
and defend every price.
Tell us your routes. FlySafe monitors them continuously, alerts you the moment an exposure crosses a repricing threshold, and hands you the cited, dated evidence to act on — no integration, no API to wire up.
Built around the AVN52 7-day reprice/cancel lever — the data that tells you when to pull it, and the audit trail to defend it afterward.
Directional estimate calibrated to publicly-reported market ranges (Lloyd’s List, Joint War Committee, Lockton). Not a bound quote. FlySafe is a data provider, not a broker or insurer.
A free account adds the cited, per-score audit trail — the exact dated sources behind this route, linkable for your file.
Get the cited audit trail — free, no card →A free directional estimate for one route — the same engine that watches your whole book. In Book Watch your routes are monitored continuously, not typed in one at a time.
If you underwrite aviation war-risk, you know this list.
- ▸Pricing on week-old PDFs while war-risk premiums change in hours.
- ▸Defending methodology to PRA / Lloyd's HHC with whatever your data vendor gave you last week.
- ▸Reading AVN52 clauses your data vendors don't understand.
- ▸Building Excel aggregations every Friday for the syndicate report.
- ▸Watching loss data lag reality by quarters.
- ▸Watching capacity withdraw in hours when an event drops.
A broker calls Friday at 5pm — premium on the DEL–AMS hull, by Monday close. Your last data PDF is Tuesday's. The route now crosses three FIRs that weren't yellow when that PDF was written.
You quote on instinct, or you miss the line.
Whether the answer was right depends on data you couldn't have. That's the gap FlySafe closes.
We watch your book. You get the signal to act — and the evidence.
You don't price a calculator — you price a book of hulls flying real routes. Book Watch monitors that book against the live engine so you're not the last to know when an exposure moves. No integration: you send the routes, we run the watch.
Send us your book — once.
Closes: rebuilding Excel route aggregations by hand every Friday. A route / hull list in any format (CSV, the binder, a spreadsheet). We load it — no API, no PAS change.
We watch it continuously and alert on movement.
Closes: capacity withdrawing in hours while you're on week-old data. Email / Telegram alert the moment a route on your book crosses a risk threshold — the AVN52 7-day window stays actionable. Weekly exposure digest for the syndicate report.
Reprice with a defensible Pack in hand.
Closes: defending the price to PRA / Lloyd's validation with "the vendor said so." Each alert and quote comes with a Defensibility Pack (below) — the dated, cited basis you put in the file.
Set it up right here — add your routes and email below and we start the watch. First exposure digest within one business day. No call, no integration.
Start Book Watch →The number is free. The evidence to defend it is the product.
A calculator gives you a figure. A validator wants to know why. The Defensibility Pack is a file-ready document behind any route or quote — the audit trail that turns "the vendor said 47" into evidence your model-validation team can actually use.
- → The score + contributing FIRs for the route
- → The dated, cited sources that moved it (EASA CZIB ref, NOTAM IDs, GDELT / ACLED events)
- → The methodology note + falsifiability criteria
Structured to map to Lloyd's IMV templates (Validation Review Signposting · Data Directory · Data Audit · Model Validation). Your validation team lifts the artifacts straight into theirs — per IMV §2.10.1, every expert judgment is documented and falsifiable.
The Pack is a paid deliverable (per-quote or bundled with Book Watch). The free calculator shows the signal classes behind a score; the Pack adds the exact dated citations and the IMV-mapped document.
What replaces "Tuesday's PDF" — three specific changes.
Defensible methodology for your syndicate's view of risk.
"A change to a third-party provider view of risk cannot be automatically accepted as representing an update to the syndicate's view of risk."— Lloyd's Internal Model Validation Guidance, January 2023
Translation: you need an audit trail that lets you validate FlySafe against your own profile. We give you exactly that — per-FIR source lineage, recalculation reproducibility, model-change documentation that maps to Lloyd's Validation Review Signposting Template. Your model adequacy story stops being "the vendor said so."
Latency that matches AVN52's actual mechanics.
AVN52 lets you cancel or reprice on 7 days' notice in the London market. That lever is only as useful as the data telling you when to pull it.
FlySafe webhooks fire on score-change with HMAC-signed payloads — same minute the model updates, not when the next PDF lands Tuesday. Per-route score lookup: 50 ms. Batch portfolio scan: configurable, scales to AVN52E sub-limit math.
Market context you can defend, not invent.
Marsh (Aug 2023) documented aviation war-hull capacity falling from ~$2B (2021) to ~$1.4B (2023) at the trough of the post-Russia hard market. That market is what your bordereau report has to defend each quarter. FlySafe gives you the per-FIR component of that picture refreshed continuously — not the index, the inputs.
What we are NOT promising
- ×100% accuracy. Calibration is ongoing. Current Brier / ECE available under NDA, methodology paper for compliance review on request.
- ×A black box. Per Lloyd's IMV §2.10.1 falsifiability requirement, our expert judgments are documented. You can pick them apart.
- ×Universal FIR coverage. 25 ML-scored in the current model, 426 mapped. The rest is roadmap.
Sources cited above: Lloyd's Internal Model Validation Guidance (January 2023), Lloyd's Model Change Guidance (June 2025), Marsh — Aviation War-Risk Coverage Presents Challenges (August 2023).
How we built it — and what we don't hide.
Where the data comes from — 15+ public sources
EASA Conflict Zone Information Bulletins · FAA SFARs + prohibitory NOTAMs · ICAO publications · national CAA NOTAMs
ADS-B with NIC/NACp position-quality reports (low-NACp = potential GPS interference signal)
UCDP (Uppsala) · ACLED · GDELT news signals · NASA FIRMS hotspots
OPSGROUP airspace warnings · published carrier deviation patterns
What we publish
- →Methodology paper — categorical signal description, scoring pipeline architecture, ENID handling. Available on request, NDA-gated for the calibration appendix.
- →Briefings — per-event analysis with sources cited, drawing on an editorial corpus of 119 curated sources, 672 principles, and 656 worked examples. See Pakistan–India 2026 · full archive.
- →Per-FIR source lineage in every API response — every score points back to the sources that moved it.
How it maps to Lloyd's IMV
Lloyd's Model Validation page lists four template categories: Validation Review Signposting · Data Directory & Data Policy Evidence · Data Audit Framework · Model Validation Evidence.
Our methodology paper structure maps directly to these on request — not because we're a syndicate, but because your validation team can lift FlySafe artifacts straight into theirs.
What we do NOT claim
- A finalized calibration. We publish Brier / ECE on a quarterly backtest cadence under NDA; methodology updates are versioned.
- Independence from your judgment. Per Lloyd's IMV §2.10.1 falsifiability requirement, every expert judgment in our model has stated criteria. You can disagree with any of them on documented grounds.
- Replacement for a primary-source review. We aggregate and score; your team retains the final view-of-risk decision.
Already have a dev team? The same engine, as an API.
Most underwriters use Book Watch — we run the monitoring for you, no code. If you'd rather pull scores straight into your PAS or pricing engine, the same data is a one-GET API. Optional, not the default.
Replace DEL→LHR with the route on the binder.
/v2/route/risk
under the hood — same response, same lineage. ~50 ms per call.
POST an array of routes. Score + premium-delta indication per route + portfolio aggregate. Excel export in your format on the response.
Subscribe to webhooks on score-change for every route in your book. HMAC-signed. Default ±15 score points or 25/50/75 boundary crossing.
$ curl https://api.flysafe.zone/v2/route/risk \
-H "X-Api-Key: $FLYSAFE_KEY" \
-G --data-urlencode "from=DEL" --data-urlencode "to=AMS"
{"score": 47, "risk_level": "moderate", "confidence": 0.82, "affected_firs": [...]}Five capabilities, framed for your workflow.
Per-route + per-FIR scoring
/v2/route/risk + /v1/firs — 0–100 risk score with source attribution, refresh 5–30 min
vs PDF vendors (quarterly bordereau lag) or rate-engine internal data (no event-driven update)
Quote war-risk routes against today's exposure, not last quarter's bordereau
Webhook score-change alerts
HMAC-signed POST on threshold crossing (default ±15 or 25/50/75 boundary)
vs daily / weekly pull cadence
Mid-policy reroute scenarios surface automatically; AVN52 7-day repricing window stays actionable
Portfolio batch + Excel export
POST array of routes → aggregated score + per-route premium-delta indication, CSV/Excel out
vs spreadsheet manual sweep every Friday
Bordereau report drafted from one API call
Source lineage in every response
Each score carries the sources that moved it (EASA bulletin reference, NOTAM ID, etc.)
vs aggregate "risk index" with opaque inputs
Maps to Lloyd's IMV §4.6 expectation that material expert judgments be traceable
Methodology + briefings on every public event
Per-event analysis with sources cited, archived at /briefings/. Methodology paper with calibration appendix (NDA)
vs proprietary scores you can't quote in a validation review
Quotable material for your internal validation pipeline and bordereau commentary
Pricing, by the book we watch.
Priced by the size of the book under monitoring + the defensibility you need — not by API request volume. A solo underwriter never outgrows a generic request tier, so the tiers scale with hulls watched instead.
- Public calculator — score + band, any route
- War-risk briefings
- — No book monitoring
- — No Defensibility Pack
- + Per-FIR source lineage
- + Excel / CSV export
- + Methodology paper (NDA)
- + 7-day audit log
- + Portfolio Risk Dashboard
- + Defensibility Pack (PDF per quote)
- + Capacity Monitor + Webhooks
- + 30-day audit log
- + AVN52 Trigger Watch
- + Monthly Pricing Briefing
- + White-label, dedicated infra
- + 365-day audit log
Sample Portfolio Risk Dashboard for a 50-hull syndicate book — live FIR scores, capacity gauges, overnight movers.
Aviation data vendors don't publish war-risk pricing or refresh cadence. Vendr's buyer-guide reports Cirium subscriptions at ~$31k/yr average with no published tier table. FlySafe's pricing is on this page.
How to start.
Start Book Watch — right on this page
Add your routes/hulls and email below. We start the watch and send alerts + weekly exposure digests — with a Defensibility Pack on any quote. First digest within one business day. No call, no integration.
Add my routes →Request the methodology paper
Categorical signal description, scoring pipeline architecture, ENID handling, falsifiability criteria per Lloyd's IMV §2.10.1. Calibration appendix NDA-gated; rest open.
Request paper →Get a sandbox API key
For teams who want to pull scores into their own systems: 5 minutes from signup to first
/v2/route/risk call. No credit card.
No commitment until you've seen it work on your own book. The free calculator above is permanent; the methodology paper is documentation, not signup.
Start Book Watch — add your routes
Paste your routes/hulls and your email — we start the watch and your first exposure digest lands within one business day. No call. (Want the methodology paper or a sandbox key instead? Say so here.)
Specific questions we get from syndicates.
Frequently Asked Questions
Common search queries answered with current status, FIR codes, and source citations.
- Do we have to integrate anything to use Book Watch?
- No. You sign up right on the page — no call. Send your routes/hulls in any format (CSV, spreadsheet, the binder) and we run the monitoring for you — alerts by email or Telegram, a weekly exposure digest, and a Defensibility Pack on any quote. First exposure digest lands within one business day. An API is available separately for teams who want to pull scores into their own systems.
- How is Book Watch priced?
- By the size of the book we watch (hulls/routes under monitoring) plus the defensibility you need — not by API request volume. A solo underwriter doing occasional quotes never outgrows a request tier, so we price on what is actually valuable: the book under watch and the audit-ready evidence. We confirm the tier from the routes you send — no call.
- How does this integrate with Subscribe / IRIS / our PAS?
- REST API + CSV/Excel export. No on-prem agent required. If your PAS exposes a webhook endpoint, score-change alerts can post there directly. Custom integration discussed at Enterprise tier.
- What happens to our model-adequacy story under Lloyd's IMV?
- FlySafe outputs include per-FIR source lineage and stated falsifiability criteria per IMV §2.10.1. The methodology paper structure is built to map to Lloyd's four template categories — Validation Review Signposting, Data Directory & Data Policy Evidence, Data Audit Framework, Model Validation Evidence. Your validation team retains ownership of the syndicate's view of risk; we provide the artifacts.
- What if FlySafe disagrees with our underwriter judgment?
- Then you're doing your job. Per Lloyd's IMV: "A change to a third-party provider view of risk cannot be automatically accepted as representing an update to the syndicate's view of risk." FlySafe is one input — the final view of risk is yours.
- How do you handle gateway-blocked sources (London QVA, Toulouse MetGate)?
- Currently out of scope. We document the gap publicly in the methodology paper. For carriers whose risk depends on those advisories, we recommend supplementing with direct subscription and flag the limitation in any pilot.
- What's your current calibration error?
- We publish Brier and ECE on a quarterly cadence against a curated event-set, under NDA. The methodology paper has the calibration appendix. Numbers vary by horizon (72h / 7d / 30d). Happy to share before pilot.
- Can we run this air-gapped / on-prem?
- Not at Pro tier. Enterprise tier supports white-label and dedicated infrastructure (EU / US / MENA regions). Air-gapped deployments are custom — discuss with sales.
- How do you handle AVN52 nuclear / UNSC-5 termination triggers?
- Both are automatic exclusions in market wording — no data vendor scores around them. Our role is to score the pre-trigger risk landscape so reprice/cancel decisions are informed before clauses activate. Per AJG's and Cambridge International & Comparative Law Quarterly analysis of AVN52E, that's where the underwriting value lives.
Built for every aviation role
The same airspace-risk engine, tailored to how your team works.