RMT Engineering Logo
Real-time floor supervision

Supervisor Console

Queue depth, agent state and SLA risk, redrawn every two seconds. And when a conversation starts to turn, the controls to listen in, whisper to the agent or take the call over sit on the same screen as the warning that told you about it.

  • Floor redrawn every two seconds
  • Listen, whisper or take over
  • Alerts before the breach
We reply within one business day. No newsletter.
2s Live floor refresh
3 Intervention modes
Wallboard Grid, list and TV views
2s Between floor and screen

See the floor as it is now, not as the report described it fifteen minutes ago

Supervisor Console is the screen a floor manager keeps open all shift. Every queue, every agent state, every live conversation: one board, redrawn on a two-second cycle. AI-handled conversations sit on that board beside the human ones, so containment is not something you go and look up in a second tool nobody has open. Then something drifts. The alert reaches you while there is still time to act on it, and the controls to intervene are already on the screen showing you the warning.

See the floor. Step in before it goes wrong.

Most floors are run off a report that went stale before anyone opened it, plus a memory of the two calls somebody happened to overhear walking past the pod, and neither of those tells you which queue is slipping at eleven this morning. This does. It shows the conversation that is turning sour while a supervisor can still do something about it.

By the numbers

Few Facts - at a glance

  • 2s

    live refresh

  • 3

    intervention modes

  • Wallboard

    grid, list and TV views

The process

How It Works

The four stages, end to end

Nothing here waits for an overnight batch. This is what happens between a conversation going wrong on the floor and a supervisor doing something about it.

  1. The floor reports in

    Queue state, agent state and every active conversation stream into the console on a two-second cycle, and the AI conversations sit on the same board as the human ones, so the containment figure updates in the same strip, on the same clock, as service level and abandon rate. One board. One operation.

  2. Thresholds watch with you

    Alert rules evaluate queue depth, wait time, abandon rate, how long an agent has been parked in one state, and live sentiment, all against thresholds you set per queue and per programme. The trick is arming the rule below the target rather than at it. A 120-second SLA gets a 90-second warning. The flag then reaches you while the target is still reachable.

  3. You step into the conversation

    Open the flagged conversation and pick one of three: monitor silently, whisper into the agent's earpiece, or take the call. All three work on voice and on chat. Each one is written to the audit trail with the time it started, the name of the supervisor who did it and the conversation it attached to, which is the difference between supervision and eavesdropping.

  4. Afterwards it becomes coaching

    The conversation is scored automatically and the transcript is searchable by phrase, so the coaching session starts from the calls that actually show the problem rather than from a general impression of how the week went. The evidence is attached before anyone sits down.

Features

Every capability you need in one module

1. The whole floor, two seconds old

Every active conversation, every agent state, every queue, redrawn every two seconds. Four views read the same floor, and the wallboard is the one built to be read from the far end of the room rather than from a desk, which is a distinction nobody notices until somebody is squinting at it. KPI strips carry service level, average handle time, abandon rate, containment and CSAT. Rankings move while you are watching them.

  • Four views of one floor
  • Grid and list for a supervisor at a desk, wallboard for the screen on the wall, manager dashboard for the week just gone
  • Service level, AHT, abandon rate, containment and CSAT in the live strip
  • Rankings that move while you watch
  • AI agents and human agents on one floor, with containment sitting in the same live strip as service level

See the 76 reports behind the live strip

Illustrative interface. Service levels, handle times and scores are sample data, not a performance claim.

2. Listen, whisper, take over

Join a live call silently and the customer hears nothing change. Whisper coaching into the agent's earpiece while the customer is still talking. Or take the conversation over, which is the rare one and ought to stay rare, because a customer who hears a second voice arrive has usually already worked out that the first one was struggling. Voice or chat, all three work the same way. Each of them lands in the audit trail with your name on it and the second it began. No exceptions to that.

  • Silent monitor — the customer hears nothing change
  • Whisper into the agent's earpiece alone, mid-turn, without the customer knowing a supervisor has joined
  • Full takeover for the calls that need rescuing rather than nudging
  • Voice and chat alike
  • Every monitor, whisper and takeover written to the audit trail with a supervisor name and a timestamp

See what the audit trail records

Illustrative interface. Names, timings and coaching text are sample data, not a performance claim.

3. Alerts fire before the breach

An SLA-at-risk alert that fires after the SLA is missed is a report, not an alert. These fire before. Sentiment-dip detection flags a live conversation going bad while it is still going, on the turn rather than in next week's sample. Rules cover queue depth, wait time, abandon rate, how long an agent has been sitting in wrap-up, and any threshold you care to write yourself.

  • SLA risk raised while there is still time to act on it
  • Sentiment dip detection on the live conversation
  • Queue depth, wait time, abandon rate, agent-state duration
  • Custom phrase match on the live transcript
  • Thresholds per queue and per programme
  • New rules run as drafts first, so you can watch what they would have fired on for a few days before you arm them and find out the first threshold was too tight

Watch it on your own floor

We will run the console against a sample of your queues and show you the alerts you would have had.

Book a demo

Illustrative interface. Thresholds, timings and rule names are sample data, not a performance claim.

4. Coaching that starts from evidence

Most coaching conversations start from the three calls somebody happened to listen to. Every interaction here is scored automatically, on 15 dimensions, so the starting point is the whole month rather than a thin sample of it. Coaching plans are generated against the dimensions an agent is genuinely weak on, with the lowest-scoring calls already attached, and calibration runs blind so that the variance analysis can tell you which of your evaluators has drifted away from the rest of them. They all drift eventually.

  • Every interaction scored on 15 dimensions, not a 2% manual sample
  • Plans generated against the dimensions that are actually weak
  • The lowest-scoring calls attached to the plan, ready to open in the session
  • Calibration with blind scoring
  • Evaluator variance, so you can see who marks soft and who marks hard before it shows up in an agent ranking
  • Schedule adherence tracked live, with deviation alerts

Inside the 15-dimension quality scoring

Illustrative interface. Scores, counts and plan text are sample data, not a performance claim.

5. Find the five calls that matter

Full-text search across every transcript you hold, voice and chat. Type the phrase. Jump to the second it was said. Filter by agent, outcome, sentiment, score, disposition or compliance flag, and the fifty thousand conversations you recorded last month come down to the five the question was actually about.

  • Full-text search across every transcript, voice and chat
  • Timestamp navigation straight to the phrase
  • Filters for agent, outcome, sentiment, score, disposition and compliance flag
  • Results open in the transcript at the exact second, so nobody scrubs through a recording hunting for the moment

See the agent desk these conversations run on

Illustrative interface. Names, dates, counts and transcript text are sample data, not a performance claim.

Use Cases

Where Supervisor Console delivers value

SLA-at-risk and queue-depth alerts

The queue that was about to miss

Harlow Utilities (illustrative)

Scenario

Billing runs a 120-second service-level target, so the wait-time rule is armed at 90. Mid-morning it fires. The console has Billing at 108 seconds and climbing, while two rows down Deliveries is comfortably under target with four agents sitting idle.

Outcome

Capacity moves across before the target is missed, rather than being explained the next morning in a meeting about why it was missed, and the rule that caught it stays on the board afterwards with the threshold it fired at still showing. Nobody has to reconstruct it.

Silent monitor and whisper coaching

A whisper instead of a transfer

Calder & Vale, a mid-market lender (illustrative)

Scenario

A customer billed for a third month on a plan she cancelled is getting louder, and after three negative turns the sentiment-dip alert puts the conversation on the supervisor's screen. The supervisor listens for one turn. Then she whispers the approved remedy across to the agent, and the customer carries on talking without a pause or a hold.

Outcome

The customer hears one continuous conversation with one person. The record afterwards shows when the monitor began, when the whisper went across and which supervisor sent it.

Full-text transcript search with timestamp jump

Five calls out of fifty thousand

Northbank Insurance (illustrative)

Scenario

Compliance wants to know which agents promised a refund without reading the disclosure first. The old answer was to sample and hope. Instead the QA lead searches the phrase across every transcript, filters to the compliance flag, and has five hits out of fifty thousand.

Outcome

The review is a short list by lunchtime instead of a week of listening, and each call ends up attached to the coaching plan the agent will actually see, opening at the timestamp where the promise was made rather than at the start of a long recording somebody then has to scrub through.

At a glance

Specification

The numbers and limits, without the sales copy

Specification for Supervisor Console
Specification Detail
Refresh rate 2 seconds
Views Grid list wallboard manager dashboard
Live KPIs Service level average handle time abandon rate containment CSAT longest wait
Interventions Silent monitor, whisper coaching, full takeover — voice and chat
Alerts SLA risk queue depth wait time abandon rate agent-state duration sentiment dip custom phrase match
Alert scope Thresholds set per queue and per programme; rules held as draft or armed
Playback Full-text transcript search, timestamp navigation, multi-criteria filtering
Quality scoring 15 dimensions, applied to every interaction rather than a sample
Coaching Auto-generated plans, calibration with blind scoring, evaluator variance analysis
Adherence Schedule adherence tracking with real-time deviation alerts
Access scope Search and playback limited to the queues and teams a supervisor is entitled to
Audit Every monitor, whisper and takeover recorded with supervisor and timestamp
FAQ

Questions,
answered

What teams ask us before they roll out Supervisor Console — how it works, what it needs from your side, and what happens when it gets something wrong

Still not sure?

Talk to a specialist and get a straight answer.

Ask our team

Yes, and both sit on the same floor in Supervisor Console rather than in two systems. The share of conversations handled by AI appears in the same KPI strip as service level and abandon rate, and when an AI-handled conversation needs a person, the handover shows up on that same board instead of in a tool only the automation team ever logs into. You are watching one operation, not a bot dashboard parked next to an agent dashboard.

A silent monitor in Supervisor Console changes nothing the customer hears, and a whisper goes into the agent's earpiece alone. Only a takeover is audible to the customer, which is rather the point of it. All three are written to the audit trail with the supervisor's name and the time the intervention started, so nothing that stays silent to the customer is also invisible to the record.

Supervisor Console needs your service-level targets and the thresholds you want warning at, set per queue and per programme. In practice that means a wait-time rule armed below the SLA, a queue-depth ceiling, an abandon-rate limit and a cap on how long an agent can sit in wrap-up. Sentiment-dip detection needs nothing beyond the transcription you already run. New rules can be held as drafts while you watch what they would have fired on, which is usually how a team finds out its first threshold was too tight.

Supervisor Console is the live surface. Quality Management supplies the interaction scores across 15 dimensions, the calibration sessions and the generated coaching plans, while Workforce Management supplies the schedule that any adherence deviation is measured against, which is why the adherence figure only appears once the rosters are in. Supervisor Console runs on its own perfectly well. The rankings and the coaching section are simply thinner without the other two behind them.

We run the platform and you choose the region the data sits in, so the recordings and transcripts Supervisor Console plays back stay in the jurisdiction you nominate. Search and playback are scoped to the queues and teams a given supervisor is entitled to see. Every playback and every intervention is recorded against the conversation itself, so a reviewer coming to it months later can see who listened to what.

Keep exploring

Additional supporting modules

Native modules that share the same record, knowledge and governance

Platform module

Agent Desk

The workspace you are watching. Live copilot, customer 360, call controls, and wrap-up drafted before the agent reaches for the keyboard.

Explore
Platform module

Quality Management

The 15 dimensions behind the rankings, plus the calibration sessions and the coaching plans they generate.

Explore
Platform module

Workforce Management

Where the adherence figure comes from — Erlang-C forecasting, schedules and intraday reoptimisation.

Explore
Connected solutions

Where Supervisor Console is used

The Solutions pages that lean on this module, and what it looks like once it is configured for a particular floor, job title or job to be done.

5 solutions built on this module

By industry

Sector floors that run on it

1

By role

Job titles that answer for it

1

By use case

Jobs it is put to work on

3
Browse every solution
Talk to a specialist

Watch a day of your own floor go past

Bring a sample of your queues, your service-level targets and a handful of recorded conversations. We'll arm the alert rules against them and show you the moments where a supervisor could have stepped in. Not a canned demo.

  • 30 minutes
  • Your queues, your thresholds, your recordings
  • No slide deck unless you want one

Book your slot

Leave your email and our team will come back to you within one business day.

or reach us directly
Call our team Email us

Your details stay private. We never share them.

This website uses cookies.

Cookies are small text files that allow us to create the best browsing experience for you on our site. By continuing to use this website or clicking "Accept & Close", you are agreeing to our use of cookies. To understand how we use cookies or how to manage them, please see our cookies policy.

Ask OptiML

Powered by RMT Engineering