HELM · Agent Console

Queue: 2 waiting · longest 0:42AHT 6:12

↺ 3 callbacks queued · next 4:30 PMSL 92%

Member IDNamePhoneEmailAddress

Answer a call to see the full context

When the HELM AI Agent (Sophie) hands you a call, the member is already identified and the whole account — subscription, connected home, the open case, and the AI's suggested next steps — lands here instantly. The customer never repeats themselves.

Supervisor — today across 3 partners

Live from the HELM SoR event stream + CCaaS stats · QM by OttoQA

92%

Service level — answered <60s

▲ 3 pts vs 7-day

26%

AI containment today

on MVP target 20–30%

2 waiting

Queue now · longest 0:42

healthy

Callbacks in queue · next 4:30 PM

callback-in-queue on

Transfer-back rate (to ISP)

watch — billing confusion

78%

FCR — 7 days

▲ 2 pts

Agents — now

Agent	State	Member · partner	Time
Alicia Romero (you)	On call	Marcus Bell · BrightSpeed	04:12
J. Chen	On call	P. Whitfield · Google Fiber	11:03
P. Ibarra	Wrap-up	— · TDS Telecom	01:20
D. Fox	Available	—	—

QM — OttoQA · 100% of calls evaluated

avg 91 / 100 this week

INT-4471-0007-03 · A. Romero · BrightSpeedflag: recording disclosure missed

INT-0094-0031-02 · J. Chen · Google Fiberflag: disposition incomplete

INT-7715-0002-02 · P. Ibarra · TDSscore 97 — exemplar

Learning loop — escalation reasons this week

Physical action required34%

Visual verification needed22%

Low AI confidence18%

Member requested human15%

KB gap11%

Every wrap-up writes an escalation reason + "could the AI have solved it?" to the SoR. This chart is the AI training backlog — the learning loop the MVP is judged on.

BPO ops checklistthe operational requirements from the 7/6 thread — and where each one lives in this console

Also tracked: recording-consent line in the IVR (decision register #7) · role-based access per partner · every write appended to the SoR audit log. Compliance/security posture rides the HELM platform controls.