Know which of your lanes just got riskier, before it is a headline.
KENNEDY watches the open feeds the world already broadcasts (armed-conflict events, ship positions, aircraft tracks, satellite fire detections), and turns four firehoses into a short, ranked, sourced list of what actually threatens the ports, routes, and suppliers you depend on. Not a raw dashboard you have to read, and not a black-box score, an alert with its evidence attached.
Put one lane under watch See a sample alert
THE PROBLEM
The signals are already public. When a conflict flares near a chokepoint, a tanker goes dark off a coast you ship through, aircraft patterns shift over a supplier region, or a fire lights up a facility, that information is broadcast in the open (ACLED logs the conflict event, AIS carries the vessel position, ADS-B carries the aircraft track, NASA FIRMS carries the thermal detection). The problem is not access, it is that nobody on your team can watch four global feeds at once and know, in the moment, which blip is a threat to your assets and which is noise. So today you either pay analyst rates for a human intelligence subscription that cannot scale to every lane, or you buy a raw feed and end up watching it yourself (which means you are not watching it at all). And as of July 26, 2026, when EU member states must have the Corporate Sustainability Due Diligence Directive transposed, continuous, evidenced supply-chain risk monitoring stops being optional and becomes a documented legal duty with a date on it.
HOW IT WORKS
KENNEDY does the step the incumbents charge for, the classify step. You register what you care about (a chokepoint, a port pair, a supplier region, a vessel), and the monitor ingests the live feeds, classifies each event as a threat relative to those specific assets, correlates multi-domain clusters (a conflict event and a fire and a vessel diversion in the same place and window are one story, not three), and dispatches a graded alert. The alert-dispatch step fails closed, so a low-confidence event never pages you as if it were confirmed, it is held and flagged instead. Every alert carries the sources it was built from, so you can check the work, not just trust the score.
WHAT YOU GET
For each asset you put under watch: - a short ranked list of events classified as threats to that specific lane, port, or supplier, not the whole world's noise, - a severity grade on each, with the low-confidence ones held and flagged rather than sent as confirmed, - correlated clusters, where a conflict, a fire, and a vessel diversion in the same place and time are surfaced as one developing situation, - the sources attached to every alert (the underlying event records), so an alert is auditable evidence, not an opinion, - delivery where you already work, on the enterprise tier, via API or webhook into your own systems.
WHO IT IS FOR
Supply-chain and logistics risk teams exposed to maritime chokepoints, cross-border routes, or single-region supplier concentration. Commodity traders, insurers and underwriters pricing geopolitical risk, and corporate security teams that today buy expensive human-analyst intelligence. It is not for teams with no physical exposure (a pure-software company has nothing here), and it is not a source of classified or paid proprietary intelligence, it works from open feeds by design.
PRICING
You start small and grow with the value, you do not sign a platform contract on day one.
- Single-lane watch, put one chokepoint or one port pair or one supplier region under watch, and see real graded alerts against your own assets first (hypothesis: around $500 per lane per month). - Asset-metered, per registered asset under continuous watch once the alerts prove useful (hypothesis: around $120 per asset per month). - Enterprise / API, the full WATCHTOWER workflow with a global risk register, custom thresholds, escalation protocols, and alert delivery into your own systems (value-metered, negotiated).
These numbers are hypotheses we are validating, not commitments. Because the monitor rides open feeds, our marginal cost stays near zero, so the ladder is about the value of the alert, not recovering a data bill. Any billing step is gated and fails closed, nothing charges without an explicit confirmation.
PROOF (how we dogfood it)
We point KENNEDY at our own supply chain. AYA's critical dependencies are its vendors, a single LLM provider (OpenAI, a deliberate single point of failure we run on purpose), a cloud host region, and our domain registrars. We register those as assets, so the same classify, correlate, and dispatch path we are selling you is the path that would page us when a geopolitical or infrastructure event threatens a vendor we cannot instantly replace. The deliverable is the Kennedy acolyte's own WATCHTOWER module, which AYA already runs internally, so eating this product is pointing an existing module at our own risk register, not standing up something new for the demo.
HONEST NOTE
Here is what is true today, and what is not. The monitor is built as composition, the feed fetchers, the threat classifier, the correlation span, and the alert dispatcher all exist on disk as verified patterns, and the deliverable workflow was PQS-scored before this page was written. What has not happened yet is a live run. Each feed needs a real API key (ACLED, AISSTREAM or AISHUB, ADSB-Exchange or adsb.lol, NASA FIRMS), and for this draft no feed was fetched, no classification was executed, no alert was dispatched, and nothing was spent. So this is a structural product that is credential-blocked, not code-blocked, and we will not claim a live monitoring result until an actual credentialed run leaves a receipt. When we show you a sample alert, it will be labeled as a sample until that run exists.
*This page is a specification. The capability it describes is not built yet, and nothing here is a claim that it runs today.*