How GAMWIT connects to what you already run
GAMWIT runs on top of the systems you already have. You do not build or replace anything. Your data connects into a pipeline that already exists, the models train on it, and the predictions come back to where your teams work. Your side is four things: requirements, a data sample, access, and feedback on the demo.
- Your side
- 4 things, 5 of 9 phases
- Live in
- 2 to 4 weeks
- Runs in
- shared cloud, your cloud, or on-premise
- Trained on
- your own players
What goes in, what comes out
What happens to your data, step by step
Four steps, all inside GAMWIT
- Connect
- GAMWIT fetches your data from wherever it already is: an SFTP upload, an S3 bucket, BigQuery, or views on your warehouse.
- Check
- Every file is validated and quality-checked before anything else reads it.
- Prepare
- The data is transformed and aggregated into the inputs the models read.
- Predict
- Each model trains on your own players, is tested on players it has not seen, then scores the live base every day. The scores go to dashboards, alerts, your warehouse and your CRM.
The nine phases, and which ones need you
Your side of it is requirements, sample data, access, and feedback on the demo. Four of the nine phases need nothing from you at all. From data access to daily predictions takes 2 to 4 weeks.
01Discovery and requirements
Prediction period, brand and the other model settings are agreed, along with what each model reads and returns.
YouShare the business requirements.
GAMWITFinalize inputs and outputs.
02Data mapping
Sample data is reviewed together, questions resolved, and the mapping to GAMWIT's API agreed. Where a standard file cannot be provided, GAMWIT builds the customization.
YouShare sample data, answer data questions, agree the mapping.
GAMWITFinalize the mapping, test the upload, build any customization.
03Data access and validation
Read access to your warehouse, or SFTP for incremental files. Quality checks run, feedback goes in, write access for prediction output is tested.
YouGrant access.
GAMWITRun the checks, incorporate feedback.
04Historical backload
Your history is loaded, checked against the raw files, and prepared for training.
YouNothing needed
GAMWITEverything.
05Model training
GAMWIT trains and tests every model on your history until the best version is chosen.
YouNothing needed
GAMWITEverything.
06Model demo
Results are summarized for your review.
YouReview the demo.
GAMWITPrepare and present it.
07Testing and feedback
Your feedback goes in and testing is finalized ahead of go-live.
YouGive feedback.
GAMWITIncorporate it, finalize testing.
08Go-live
The models run in production and daily predictions begin.
YouNothing needed
GAMWITEverything.
09Ongoing
Model performance is monitored continuously, and models are retrained when player behavior changes.
YouNothing needed
GAMWITEverything.
What you bring, and what comes back
You bring
- 90 days of history
- Most models train on 90 days; Responsible Gaming asks for 180. GAMWIT keeps 1.5 years once you are live.
- The fields you already have
- Betting activity, deposits, withdrawals, bonuses, NGR and GGR carry most models. Each model page lists exactly what it reads.
- No personal data
- The models work on behavior. No personal or identifiable player information is stored, and GDPR and Data Processing Agreements are signed as standard.
What comes back
- Daily predictions
- A scored list from every model in scope, with its results and a drift view, so you can see how it is doing.
- Dashboards, alerts and self-service BI
- On the same data, for every team.
- Files into your systems
- Predictions written back to your warehouse and to any CRM that takes a file or an API push, so your own tools act on them.
Runs where you need it
Shared cloud, your own private cloud, or on-premise. Nothing you already run is replaced.
See the connection mapped to your own systems.
Bring the list of what you run. The walkthrough maps it to the phases above.