Semantic HTML, CSS, and vanilla JavaScript served through the portfolio’s S3 and CloudFront hosting.
Customer lifecycle intelligence
See the evidence behind renewal-risk priority.
Move from portfolio patterns to account evidence while keeping observed signals, model interpretation, and human judgment separate.
- Live read-only API
- Explainable ranking
- Human review boundary
- Live read-only API
- Explainable ranking
- Human review boundary
Portfolio pulse
Where should the team investigate?
Loading synthetic reference data…
Account investigation
Prioritized evidence review
| Account | Start date | Region | Segment | Sector | Touchpoints | Risk | Percentile | Adoption | Cases |
|---|
How this reference build works
Live queries, precomputed synthetic scores.
GET-only API Gateway routes backed by one Python 3.12 Lambda function.
Bundled read-only SQLite data with 4,000 fictional accounts. No customer or employer data.
Precomputed reference-model scores support exploration; the demo performs no live training, inference, or automated customer action.
Method note: This reference build is separate from the frozen Candidate 10 experiment documented in the case study. Their metrics are intentionally not combined.