Calling Desk
19Operational console for the calling team. Each task is routed to the AI voice agent or to a human collector based on the AI-vs-human suitability model — every routing decision is explainable, every cost difference is visible.
AI queue
7
@ $0.18 / call
Human queue
7
@ $4.20 / call
Live now
2
AI + human
AI → Human escalations · 24h
2
hardship · ≥2 RPC fail
PTPs captured today
1
$1115 expected recovery
AI cost vs human cost
$1.80 / $37.80
cumulative today · 5% AI
Cohorted for the AI voice agent. Cheap and high-volume.
| Customer · account | Mode | Step | A × W | Reason | RPC · PTP | EV / cost | Window | Assignee | Status | |
|---|---|---|---|---|---|---|---|---|---|---|
| Aisha Khan PFNA-2025-1117 · 12d DPD · US | AI | 1/5 · Soft AI | HardshipA20 · W73 | Active PTP due 26 Apr · light reminder cycle 24h before due-date. | 62% · 71% P(RPC) · P(PTP-kept) | $18 cost $0.18 | 10:00–12:00 CT | AI · Cara v3.2 | Queued | |
| Rohan Mehta PFNA-2025-0987 · 18d DPD · UK | AI | 1/5 · Soft AI | RecoveryA2 · W28 | Self-cure score 0.78 — soft nudge before 30 DPD bucket roll. Vulnerability flag | 55% · 66% P(RPC) · P(PTP-kept) | $14 cost $0.18 | 12:00–14:00 CT | AI · Cara v3.2 | Queued | |
| Niran Phasuk SYN-Niran-9 · 9d DPD · TH | AI | 1/5 · Soft AI | HardshipA39 · W86 | Bucket-roll model 0.34 · WhatsApp opt-out · voice-AI fallback. | 48% · 59% P(RPC) · P(PTP-kept) | $22 cost $0.18 | 16:00–18:00 ICT | AI · Cara TH v2.4 | Dialing | |
| Pim Boonmee SYN-Pim-22 · 22d DPD · TH | AI | 2/5 · AI follow-up | HardshipA39 · W73 | PTP captured by AI on 22 Apr — confirm before due-date 28 Apr. | 51% · 69% P(RPC) · P(PTP-kept) | $20 cost $0.18 | 10:00–12:00 ICT | AI · Cara TH v2.4 | Queued | |
| Apirak Chaiyakit SYN-Apirak-26 · 26d DPD · TH | AI | 2/5 · AI follow-up | HardshipA32 · W86 | Self-cure stalled · soft AI follow-up before manual escalation. | 44% · 52% P(RPC) · P(PTP-kept) | $16 cost $0.18 | 14:00–16:00 ICT | AI · Cara TH v2.4 | Queued | |
| Mei Chen SYN-Mei-14 · 14d DPD · ID | AI | 1/5 · Soft AI | HardshipA27 · W86 | Inbound chat handed back to outbound voice for ID confirmation. | 58% · 62% P(RPC) · P(PTP-kept) | $20 cost $0.18 | 13:00–15:00 WIB | AI · Cara ID v1.6 | Queued | |
| Kanya Wongsuwan SYN-Kanya-0 · 0d DPD · TH | AI | 1/5 · Soft AI | HardshipA50 · W76 | Pre-due 7-day reminder · WhatsApp + voice-AI hybrid cohort. | 61% · 74% P(RPC) · P(PTP-kept) | $24 cost $0.18 | 17:30–19:30 ICT | AI · Cara TH v2.4 | Queued |
Routing rationale · sidecar
Rohan Mehta → AI (98% confidence)
Ability × WillingnessRecovery
A 2·W 28
Recovery candidate · low ability and low willingness. Prepare repo / legal pathway with full evidence pack and supervisor review.
- Vulnerability flag+0.32
- DPD severity+0.21
- Cost asymmetry (human $4.20 vs AI $0.18)-0.18
- Broken-promise pattern+0.14
Policy gates
- AI mode disabled if vulnerability flag is active
- AI mode disabled if jurisdiction does not permit AI-driven collection at this DPD
- Human mode requires open contact window + 7-in-7 cap headroom
Calling-only escalation ladder
Lowest-friction first · cheapest channel that works
AI handles the bottom of the ladder where contact-windows are open and warmth is not required. Human seats are reserved for negotiation, hardship and high-DPD warmth. Field visits and legal escalation sit above this ladder — see Field Ops and Recovery.
Step 1
Soft AI nudge
AI · Cara
cost$0.18
RPC lift+11%
Step 2
AI follow-up
AI · Cara
cost$0.18
RPC lift+18%
Step 3
Human collector
Phone team
cost$4.20
RPC lift+34%
Step 4
Senior human
Senior negotiator
cost$6.80
RPC lift+42%
Step 5
Pre-NOI / legal
Recovery officer
cost$11.20
RPC lift+58% (asset)