UMAR AI call quality assurance
Review every call, not the handful your QA team has time to listen to.
UMAR transcribes your recorded calls, scores each agent against your own QA scorecard and puts the results in front of your QA managers. If your recordings cannot leave the building, UMAR runs inside your network.
Delivered for Special Communications Organization (SCO).
Call review view. Illustrative data.
- Built for
- Contact centres and telecoms operators
- Result
- About 10× QA throughput at a similar operating cost
- Deployment
- Cloud or on-premise, including air-gapped networks
- Scoring
- Your own QA scorecard
The problem
A person can only listen to so many calls in a day.
Manual QA means a manager plays a recording, fills in a scorecard and moves to the next one. However good the team is, most calls are never heard.
Most calls go unreviewed
Quality is judged on a small sample, so recurring problems and strong performers are both easy to miss.
Scores depend on who is scoring
Two managers can mark the same call differently, and agents notice.
More coverage means more headcount
The only way to review more calls by hand is to hire more reviewers.
Benefits
More than more calls reviewed.
Throughput is the headline number. These are the other things a QA team gets from UMAR.
10×
More calls reviewed
The same QA team reviews around ten times as many calls, so quality is judged on far more than a small sample.
Lower cost per call
Cheaper QA
Ten times the coverage for a similar running cost. More calls checked without hiring more reviewers.
Same rules
Consistent scoring
Every agent and every call is marked against the same scorecard, whoever would have been scoring it.
Every score explained
Scores you can check
Each score shows the part of the conversation it was based on, so managers can see why a criterion was marked down.
By agent and team
Coaching built in
Results build into a picture of each agent over time, ready for coaching conversations.
In your network
Private by design
UMAR can run on your own servers, including air-gapped networks, so recordings never have to leave the building.
How it works
From a recording to a score your managers can act on.
Calls are collected
UMAR picks up recordings from your telephony or call-recording system.
Speech becomes text
Each call is transcribed automatically.
AI scores the call
The transcript is checked against every criterion on your scorecard.
Managers review
QA managers check the scores and deal with flagged calls.
Results reach the dashboard
Scores are reported by agent and team, ready for coaching.
Your scorecard, applied the same way every time
UMAR marks calls the way your QA team already does.
We set UMAR up with the criteria your managers use today. Every call is judged against the same standard, and each score comes with the part of the conversation it was based on.
- Consistent. The same rules are applied to every agent and every call.
- Explainable. Managers can see why a criterion was marked down.
- Tied to the agent. Each call is matched to the agent's ID, so results build into a picture of each person over time.
- Focused on agents. UMAR evaluates how the call was handled. It does not profile your customers.
Call review view. Illustrative data.
On-premise deployment
Your recordings never have to leave your network.
Most AI QA tools need your calls uploaded to someone else's cloud. For telecoms operators, government bodies and banks, that is often not allowed. UMAR was built to run on your own servers.
Runs on your hardware
Transcription and scoring run on GPU servers inside your data centre, including networks with no internet connection.
Audio stays where it is
UMAR reads recordings over your local network. It does not keep its own copy of the audio.
Scheduled, not always on
Calls can be processed in batches, for example weekly, so hardware is not running around the clock.
Cloud if you prefer
Where data rules allow, UMAR can be deployed in the cloud with less to set up.
10×the QA throughput, at a similar operating cost
What happened at SCO.
Special Communications Organization had two QA managers who could listen to only a small share of calls. With UMAR, the same team reviews roughly ten times as many, using the same scorecard, without a matching rise in cost.
Read the UMAR case studyQuestions we get asked
What buyers usually ask before a demo. Ask us directly.
What is UMAR?
UMAR is AI call quality assurance software. It transcribes recorded customer service calls, scores each call against your QA scorecard and presents the results to your QA managers in a dashboard.
Does UMAR replace QA managers?
No. UMAR does the listening and first-pass scoring so that managers can cover far more calls. Managers review the scores, handle exceptions and coach agents.
Can UMAR run on our own servers?
Yes. UMAR can be deployed in the cloud or on-premise, including inside air-gapped networks. In an on-premise deployment your call recordings never leave your network.
Can we use our own QA scorecard?
Yes. UMAR scores against the criteria you already use, and the scorecard is configured with your QA team during setup.
Does UMAR analyse customer sentiment?
No. UMAR is built to evaluate agent performance against a scorecard. It does not profile customers or report on customer mood.
See UMAR score a call.
In a 30-minute demo we show a call going from recording to scorecard, and talk through your call volumes, your scorecard and whether you need cloud or on-premise.
- For heads of customer service, QA leads and contact centre managers
- Suited to telecoms, utilities, banking and public services
- Want AI built for a different process? See our AI development services