Call ingestion
Created the intake flow needed to bring call recordings into a repeatable QA analysis process.
AI Solutions case study
UMAR: AI-Assisted Call Quality Assurance for SCO
NexAura delivered UMAR for Special Communications Organization (SCO) to increase quality-assurance throughput for call reviews while keeping structured QA criteria and human oversight in the process.
Project summary
A completed AI-assisted QA solution for higher call-review throughput.
SCO's quality-assurance process relied heavily on manual call review. Two QA managers could assess only a small sample of the total call volume, limiting coverage and visibility into wider call-handling performance.
UMAR was delivered to substantially increase QA throughput without a proportionate increase in operating cost, using speech-to-text technology, large language models, structured QA rubrics, automated analysis workflows and human review.
- Client
- Special Communications Organization (SCO)
- Status
- Completed
- Service area
- AI Solutions
- Verified outcome
- Approximately 10× higher QA throughput at a similar operating cost.
The manual QA challenge
Manual review limited how much call-handling performance SCO could see.
With QA dependent on manual listening and scoring, review capacity was constrained by the time available to the two QA managers. A small reviewed sample made it harder to compare calls consistently or spot recurring issues across the wider operation.
- QA coverage was limited to a small proportion of total calls.
- Evaluation consistency depended on manual review time and attention.
- Wider call-handling patterns were difficult to assess at scale.
- Increasing review volume manually would have added operating pressure.
What NexAura delivered
A call-to-insight QA workflow with structured scoring and review built in.
NexAura's work covered the full path from call ingestion to a custom reporting dashboard, while keeping the technology description public, generic and focused on operational value.
Audio transcription
Used speech-to-text technology to convert call audio into text that could be evaluated against agreed criteria.
QA rubric design
Structured the evaluation criteria so calls could be assessed more consistently across the same quality standards.
LLM-assisted analysis
Applied large language models within an automated analysis workflow to review call transcripts against the rubric.
Structured scoring
Produced structured scoring and evaluation outputs designed for QA manager review, comparison and follow-up.
Review and exceptions
Included human review and exception handling so the workflow supported oversight rather than replacing judgement.
QA visibility dashboard
Delivered a custom reporting dashboard to help managers see reviewed calls, scoring outputs and recurring QA themes.
Completed solution delivery
Delivered the completed UMAR solution for SCO's call quality-assurance workflow.
Call-to-insight workflow visual
From recorded calls to structured QA insight.
The workflow connects call ingestion, transcription, rubric-based LLM analysis, QA review and dashboard visibility without representing a private product screen.
-
Call ingestion
Call recordings enter a repeatable analysis path.
-
Transcription
Speech-to-text technology prepares readable call text.
-
Rubric-based LLM analysis
Large language models evaluate transcripts against structured QA rubrics.
-
QA review
QA managers review outputs and handle exceptions.
-
Dashboard and insights
A custom reporting dashboard improves visibility across reviewed calls.
Call ingestion → transcription → rubric-based LLM analysis → QA review → dashboard and insights
QA rubric and human oversight
Structured criteria improved consistency while keeping people in control.
UMAR was designed around defined QA criteria, not open-ended analysis. The automated workflow supports the QA managers by applying the same rubric to a much larger review set and surfacing outputs for human review.
Human oversight remains visible in the workflow through review steps and exception handling, helping SCO retain judgement and context while expanding review capacity.
Structured QA rubrics
Evaluation standards were represented as clear criteria for repeatable assessment.
Automated analysis workflows
Transcripts could be reviewed at scale against the agreed rubric.
QA manager oversight
Human review remained part of the process before outputs were used operationally.
Exception handling
Calls needing closer attention could be surfaced for additional review.
Verified outcome
Approximately 10× higher QA throughput at a similar operating cost.
The confirmed outcome was a substantial increase in QA throughput without a proportionate increase in operating cost.
UMAR enabled SCO to assess a much larger proportion of calls while retaining structured QA criteria, dashboard visibility and human oversight.
Confirmed result
Approx. 10×
higher QA throughput at a similar operating cost.