Open ecosystem
Connect what golf already measures well.
swyng connects golf, health and environmental data with LLMs and visual AI. Providers are integration candidates, not claimed partners.
Planned
AI Golf Brain
SWYNG nutzt starke Datenquellen und führt sie in einer erklärbaren Intelligenzschicht zusammen.
Golf and wearable data
Garmin, Apple Health and Health Connect, WHOOP and Oura are relevant candidates for activity, load and recovery. TrackMan, FlightScope, Shot Scope, Arccos, SkyTrak and other launch-monitor, shot-tracking and simulator systems are evaluated by API access, licence and data quality.
- Garmin & wearables
- TrackMan & FlightScope
- Shot Scope & Arccos
- Simulators & launch monitors
LLM and agent layer
An interchangeable multi-model layer is planned for dialogue, explanations, planning and tool use. OpenAI, Google Gemini, Anthropic and European models can be routed by task, cost, latency, privacy and evaluation quality. The model is not the product: SWYNG’s golf context, rules, guardrails and memory form the defensible AI Golf Brain layer.
- Model routing
- Golf RAG
- Tool calling
- Evaluation & guardrails
Visual AI and swing intelligence
On-device pose estimation, club and ball tracking, swing-phase detection and video segmentation form the visual-analysis stack. MediaPipe, Apple Vision/Core ML, ML Kit and specialised proprietary models are technical building blocks; swing metrics are only presented as reliable after validation.
- Pose estimation
- Phase detection
- Club & ball tracking
- On-device inference
Course, weather and golf infrastructure
Maps, weather, course data, handicap, tournaments and club systems are handled integration-first. Candidates include Google Maps or Mapbox, weather providers, Golf Genius, GHIN, DGV and R&A-related data routes, depending on region, contract and permitted use case.
- Maps & weather
- Course data
- Handicap & rules
- Club & tournament APIs
A quality contract for every data point
Source, time, unit, method, confidence, licence, region and provenance belong in the data model. SWYNG shows whether a statement was measured, calculated, user-entered or AI-derived.
Buy, integrate or build
We buy commodity data, integrate strong specialists and build identity, context graph, golf memory, decisioning and the differentiating intelligence layer. Every integration passes security, privacy, cost and exit-risk reviews.