POST /v1/scoreSlot detail
Opening a slot in the Sealect operator view fires /v1/score. Score, status, confidence, hard gates and data freshness sit on the slot, next to the booking, not on a separate chart.
Relationship disclosure
Sealect and Goable are separate products built by the same founder. Sealect is used as a reference integration to test Goable's public API in a real booking workflow. It receives no preferential production API access, no special scoring logic and no separate service-level terms. Being founder-linked is stated plainly here so a reader never mistakes it for an independent, arm's-length customer validation.
01
What Sealect is
A booking OS for water-sports schools
Sealect Business runs water-sports schools: manual and online bookings, activities and offers, recurring slots, team roles, Stripe payments, CRM and analytics. Sealect.app is the separate consumer surface for discovering and booking water activities, with geolocated search.
That makes the relationship with Goable straightforward. Sealect owns booking, operator workflow, and customer and slot data. Goable provides weather-to-suitability decision support and score provenance. An operator uses it to weigh whether to open, hold, move or propose an alternative for a slot, without Goable becoming the booking system.
02
Why this workflow
It is where weather actually matters
Sealect manages the exact operational moments in which weather changes a decision: an operator creates activities and slots, reviews the coming week, communicates changes, and records what occurred. That gives Goable a place to test whether score context is understandable and actionable in a real scheduling workflow, without turning Goable into a booking system. The test is comprehension and fit, not a performance claim.
03
Integration scope
Phase 1: score, planner, structured outcomes
The first phase wires three surfaces that map onto objects Sealect already has: the slot, the weekly calendar, and the session close. Recovery and alternatives are deliberately deferred to a later phase.
POST /v1/score/seriesWeekly planner
Hourly series across the next seven days for the week view: a discreet colour overlay per slot with hour-level drill-down. Cached aggressively, since the planner is the most-revisited surface.
POST /v1/score/:id/outcomeStructured session close
When a session closes, the tenant POSTs back with a reason_category, not free text. That structured cause is what lets calibration separate a weather decision from a staff, payment or demand cancellation.
04
Honest scope
What this proves, and what it cannot
The page can show
- The workflow is a natural fit
Sealect manages the exact objects weather acts on: activities, recurring slots, staff, bookings, cancellations, outcomes.
- The endpoints integrate into a booking UI
Score, series and structured outcomes sit inside a real operator surface, not a demo harness.
- Score and provenance are useful to show
An operator can read the verdict, the confidence and the reason a session did or did not run.
- Goable stays an API layer
Sealect remains the workflow product. Goable never becomes the booking system.
It cannot yet show
- Accuracy or commercial uplift
One founder-linked integration is not an independent measure of forecast skill or booking conversion.
- Safety or fitness for every operator
Suitability is decision support. The operator owns every go or no-go call.
- Field-calibration effectiveness
Whether local outcomes improve the curve is measured against a holdout, not assumed.
- Market demand or willingness to pay
A reference integration validates the workflow, not the market.
05
Data and calibration
A structured outcome, and a gated loop
A cancellation can come from weather, staff, payment, equipment, customer or demand. So "ran, cancelled, no-show" is not enough to calibrate suitability. At close, Sealect records a structured reason_category. Only weather and safety outcomes count as evidence for or against the forecast score; the rest are recorded as business facts and excluded from calibration.
Ran as plannedfeeds calibration
Moved, weather-relatedfeeds calibration
Cancelled, weather-relatedfeeds calibration
Safety-related overridefeeds calibration
Cancelled, operationalexcluded
Cancelled, customer or demandexcluded
No-showexcluded
Unknown or not recordedexcluded
1
Score
An operator opens a slot. /v1/score returns a verdict, confidence and hard gates, audit-logged.
2
Session
They act on it: the activity runs as planned, is moved, or is cancelled.
3
Structured outcome
At close, the tenant records what happened and, crucially, why: weather, operational, demand, safety.
4
Evidence-gated calibration
Only weather and safety outcomes feed the local curve, and only after holdout, validation and rollback checks pass.
Outcome ingestion
Daily, where operators record a structured outcome. Ingestion is for quality checks and monitoring, not an automatic nightly refit.
Calibration
Evidence-gated. A candidate update is measured against a holdout baseline and is promoted only after validation and rollback checks. It can be rejected or rolled back.
Retention and recall
Outcomes are kept one year by default; longer only under a written retention schedule. A batch later found mislabelled can be recalled (voided) so it stops influencing calibration, without a destructive delete.
06
Fairness commitments
Production parity, collaboration declared
Same public production API
Sealect calls the same public /v1 endpoints, with the same scoring logic and the same production gates, as any external customer. There is no internal API and no preferential cell.
No preferential scoring or availability terms
Same plan limits and the same service commitments offered to comparable booking-platform customers. The two teams may collaborate on test environments, implementation feedback and issue reproduction; that is a support relationship, not a service privilege.
Comparable economics
If deployed in production, Sealect uses the public plan and usage model applicable to comparable booking-platform customers. Any pilot work is separately scoped and does not change production scoring or availability.
Operator feedback, close to the product
Because the same founder builds both, rough edges surface in days: latency, an awkward field, a confusing verdict, a provider that failed. The engine evolves to that feedback, without special-casing Sealect.
07
Later phase
Recovery options, only where coverage allows
/v1/recommend-spot is deferred on purpose. Sealect works with individual schools and their offers, not a shared, verified catalogue of operational alternatives, so proposing a nearby option risks suggesting a competitor, an unverified location or unreal availability. Where availability, catalogue coverage and operator rules permit, a later phase may surface a suitable alternative slot, activity or location that the same operator already controls.
08
For booking platforms
Bring your own workflow
Sealect is one booking workflow. Any booking platform can integrate Goable the same way, on the same public API, and hold it to the same discipline.