hemanthsai-d/AI Decision Platform — Customer Support Triage
AI decision engine for customer support triage with confidence scoring, PII redaction, and policy gates.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| issue_text | Customer support message to triage | yes | — |
| customer_tier | Customer tier (basic, premium, enterprise) | no | basic |
| channel | Support channel (chat, email, phone, web) | no | chat |
| confidence_threshold | Minimum confidence to auto-respond (0.0–1.0) | no | 0.6 |
| pii_redaction | Enable PII redaction (true/false) | no | true |
| injection_detection | Enable prompt injection detection (true/false) | no | true |
| embedding_backend | Embedding backend (sentence-transformer, none) | no | sentence-transformer |
| model_backend | LLM backend (ollama, none). Set to none for classification-only mode. | no | none |
| output_format | Output format (json, summary) | no | json |
Outputs
| name | description |
|---|---|
| decision | Triage decision: auto_respond, abstain, or escalate |
| intent | Classified intent category (one of 30 categories) |
| confidence | Final confidence score (0.0–1.0) |
| confidence_signals | JSON object with individual signal scores |
| pii_detected | Whether PII was detected and redacted (true/false) |
| injection_detected | Whether prompt injection was detected (true/false) |
| policy_override | Policy override applied, if any (null or override reason) |
| result_json | Full JSON result object |