Can I trust the forecast here today?
Senlay compares forecasts with real observations and tells you when they stop agreeing.
Personal spot watcher for watersports
Stop checking the forecast. Senlay watches your spots for you.
Tell Senlay where and how you ride — kite, wing, windsurf, or surf. It monitors forecasts, real observations, and local spot patterns, then tells you when something meaningful changes. You don't have to open the app for it to keep working.
The old way / the Senlay way
Configure once. Senlay watches. You get notified when it matters.
The old way: check Windy, check another model, check the station, check again, ask a friend, check again, guess. The Senlay way: tell it where you ride and what conditions work for you — then stop checking.
Is this spot behaving normally?
Every spot has its own rhythm — thermals, tides, direction shifts. Senlay learns it.
Is it time for me to go?
A persistent Watch checks your conditions continuously and notifies you only when something changes.
How Senlay thinks
Forecasts + real observations + local spot history + your conditions.
Senlay watches continuously. When something meaningful changes, it turns that into an event — and notifies you.
Displays vs. actuates
A weather API answers a person. Senlay answers a system.
When software shows a number to a human, the human applies their own judgement. When software acts on a number, nobody does. Senlay supplies the missing scepticism: provenance, freshness, confidence, source disagreement, and an explicit statement of what the system does not know.
Why Senlay exists
Agents should not guess when the world can be measured.
Training data can describe a place. A generic API can return a number. Senlay adds the missing trust layer: what was observed, where it came from, when it was measured, whether sources agree, and how much the result should influence a real-world decision.
Bring fragmented signals together.
Hardware sensors, public networks, forecast providers, satellite events, terrain, marine feeds, GPS tracks, and customer IoT devices enter through one integration layer.
Attach evidence to every answer.
Observations and predictions stay distinct. Freshness, proximity, confidence, source health, and model-versus-reality disagreement remain visible instead of being hidden.
Return context an agent can act on.
Field-specific rules turn raw readings into structured context, risk events, and plain-language uncertainty without pretending the API is a final safety authority.
Local observation memory
The API gets better at a place the more it is asked about it.
Eligible requests retain compact observation snapshots for that coordinate. When enough hardware-backed history exists, Senlay can describe recent local rhythm; otherwise it says that the pattern is not established.
Every request is remembered
Each call stores an observation snapshot for its coordinate, so the record of a place deepens every time it is queried.
Observed local pattern
When sufficient hardware-backed history exists: typical daily low and peak, ramp start time, and recurring gust factor. Model history alone is not presented as a learned local pattern.
Local history adds context
Eligible requests can retain compact local snapshots and compare them with current sensor/model evidence. The historical layer is descriptive today; its effect on decision quality is being evaluated rather than inferred from request volume.
Explicit uncertainty
It tells you what it does not know.
When validated bathymetry is unavailable, Senlay disables seafloor and shallow-water hazard inference and says so in the response. Elevation is never presented as ocean depth. Official alert feeds returning nothing are reported as “no events returned”, never as a safety clearance.
Systems that interpolate across their gaps are the ones that get people hurt. In a safety-critical workflow, an inferred value presented as a measurement is worse than no value.
Contribute or pay
Coverage comes from participants. Revenue comes from systems that act.
The two sides do not compete for the same budget. Contributors pay in data and receive the network; consumers pay in cash because they will never install a weather station.
Attach a sensor, get the whole network.
A weather station, an anemometer, a tide gauge, a marina, a kite station, a farm probe. Attach it directly and receive verified reasoning across every instrument around you, not just your own readings. No cash, no token.
Bring a network you already run.
Operators with existing sensor fleets connect them and gain the verification layer plus coverage they do not own. The fleet becomes more valuable because it is reasoned over, not merely read.
Systems that act, pay cash.
Drone and BVLOS operators, robotics fleets, lifting and construction, parametric insurers, marine operations. They need the answer, not the network, and they are the revenue.
One core, many fields
A horizontal sensing layer with domain-aware outputs.
The evidence contract stays consistent while the sources, rules, and decision context adapt to each physical workflow.
Autonomous and drone operations
Wind, gusts, precipitation, visibility, terrain exposure, airspace weather, and go/no-go context. BVLOS flight requires a documented pre-flight risk assessment; a provenance trail is a regulatory artifact, not a feature.
Marine and water safety
Waves, tides, current, buoy observations, wind direction, coastal geometry, GPS drift, and rescue-margin context — with SmartSurf as the proving ground.
Lifting and construction
Crane wind limits are a legal threshold. Whether one was exceeded is settled by source and timestamp, not by a nearby forecast.
Parametric insurance
A payout triggered by an environmental threshold must be both provable and contestable. Evidence with provenance and freshness is the record that settles it.
Agriculture and field work
Heat, humidity, evapotranspiration, frost, precipitation, and spray-window conditions, using the ET₀ and VPD calculators already in the modifier engine.
Environmental monitoring
Air quality, fire and hazard feeds, satellite events, earthquakes, volcanoes, and source freshness.
Connected products and IoT
Private station networks, trackers, device telemetry, customer sensors, webhooks, and normalized evidence records.
AI assistants and workflows
One fast-growing segment of the same layer: agent-ready context strings, versioned JSON, API-key access, model integrations, and auditable risk events.
How it works
From unreliable upstreams to explainable agent context.
Senlay is an end-to-end production system, not a static concept page. It handles provider integrations, normalization, trust logic, authentication, persistence, API delivery, and Linux deployment behind a public HTTPS service.
Source adapters
Connect weather, marine, aviation, environmental, satellite, and private-device APIs with bounded timeouts and explicit licensing controls.
Evidence contracts
Convert incompatible payloads into versioned observations, predictions, static context, provenance, and source-health records.
Context and risk
Fuse local evidence, terrain modifiers, fallback policy, domain rules, and disagreement into machine- and human-readable outputs.
Production delivery
Serve through Node/Express, SQLite, Nginx, PM2 or Docker with secrets isolation, health checks, logs, backups, and deployment runbooks.
Engineering proof
Built to be diagnosed, supported, and deployed.
The project demonstrates the work between “API idea” and “reliable service”: tracing failures across integrations, validating responses, managing Linux runtime state, protecting credentials, documenting boundaries, and communicating what the system knows and does not know.
API troubleshooting
Authentication, validation, rate limits, upstream errors, normalized responses, OpenAPI contracts, and reproducible test requests.
Linux / VPS operations
Nginx reverse proxying, PM2 processes, environment configuration, filesystem permissions, runtime data, smoke checks, and recovery steps.
Integration resilience
Timeouts, caching, singleflight requests, graceful degradation, source-health tracking, and transparent fallbacks when providers fail.
Technical communication
Public docs, agent-readable manifests, deployment guides, decision boundaries, provenance, and clear explanations for developers and operators.
First real-world proof
SmartSurf is where verification meets water.
Senlay grew from a practical problem: outdoor decisions depend on local reality, while models and nearby sensors can disagree. SmartSurf applies the same evidence engine to riders, boards, schools, GPS trackers, weather, marine conditions, and water-safety workflows.
Phone-first safety workflow
Sessions, GPS location, emergency contacts, local conditions, alerts, and operational history.
Tracker and sensor integration
Waterproof GPS/SOS devices and private telemetry add live rider, board, and fleet signals to environmental context.
From prototype to service
A working vertical that connects physical hardware, backend APIs, risk logic, dashboards, documentation, and production infrastructure.
Build with Senlay
Give your system a way to check the world.
A free key is one POST with one field — no password, no card. Register an agent, then inspect the evidence contract or pull live context through the documented API. The free tier is designed for prototypes and integration testing.