What's new in Call Coach IQ

A new AI Outcomes page shows the trend behind your quality, compliance, customer satisfaction and coaching results month by month, plus a maturity score that says what would move you up. Survey CSAT and the AI’s own per-call read of satisfaction are drawn side by side and never blended, so you can see how closely one tracks the other. Teams running the AI voice agent also see how many calls it resolved, how many it escalated, and the cost per resolved call. Available on every plan.

AI Semantic Search on Call History now searches every passage of every transcript, not just a summary of the call. Each match shows the sentence it matched on and, where the transcript is timestamped, one click opens the recording at that moment. Calls where the relevant part came fifty minutes in, which search could not reach before, now turn up. Existing calls have been indexed too.

Team KPIs are now built from daily summaries instead of re-reading thirty days of calls on every visit. Loads that used to take several seconds, and occasionally minutes, now come back almost immediately, and the slowest page-load requests are kept warm so the first visit of the day is not the slow one.

When a customer called a colleague first, Search History now flags any promise that is still open. Click the badge to see the promise itself, when it is due, and the line of the call it came from. If you need more, send the agent who took that call a quick question from the same screen; it lands in their notifications with a link to the call.

Resolving a flagged call now records why: a false positive, a real issue that has been fixed, and so on. Business Insights adds these up so you can see whether your team is clearing mostly false alarms or mostly real problems, and click any bar to open the calls behind it. Audits resolved before this change can have a reason added without losing the original reviewer or date.

The lenient / balanced / strict scoring setting now shows how your managers have corrected AI scores over time, both per metric and company-wide. If one metric keeps getting marked up or down, you will see that the rubric wording is the thing to change rather than the setting that moves every metric at once.

Call detail now shows a banner when the transcription is likely wrong, and says whether the words or the speaker labels are the problem. The banner is explicit that this is about the transcript, not the agent. Confidence is also tracked per phone provider, so a provider whose recordings consistently transcribe badly shows up.

A call the AI could not score is no longer marked Reviewed, and a metric the AI could not match is left unscored instead of defaulting to 75. Close rate no longer counts calls that could never have produced a sale. Calls left stuck in Pending Review are now picked up and finished automatically. A custom-data CSV whose column headers had trailing spaces no longer imports those columns blank.

When a critical compliance failure caps a call’s overall score, the call now says so. Before, the metric sliders could average 85 beside an overall score of 60, with nothing on screen to explain why the overall would not move.

For teams connected to Calix, opening a customer on Customer Journey now shows their line and equipment: whether the gateway is reachable, mesh units, Wi-Fi settings, the plan they pay for against what the line can carry, and speed-test history on the timeline. It suggests a technician visit only when a fault on the line and the customer’s contact pattern agree. Live IQ gets a plain-language card for the caller’s connection.

An escalation can now go to a department or queue instead of a named person, each destination can keep its own hours, and admins describe what each one handles. Whether anyone is available decides between a warm transfer with a spoken handover and a direct transfer; the call always reaches your own queue, where it is recorded and coached. The settings page is regrouped into six sections and shows what the agent can do with each connected integration.

Automatic recording polling can now run every 30 minutes, so calls reach coaching sooner. Options slower than every two hours have been retired, and any company still set to one of them now polls every two hours.

The AI voice agent is answering live inbound calls end to end. It greets the caller, verifies who it is speaking to before releasing account detail, answers from your own knowledge base rather than guessing, checks service availability at an address, takes a callback request, and warm-transfers to a person or a queue with a spoken handover so the human knows the story before they say hello. Every call opens with an AI disclosure your company signs off word for word.

A new Live IQ board under Dashboards shows every AI call happening right now, side by side: what the caller heard, what the agent actually looked up to say it, and what the call is costing. Supervisors can step in and take a call over at any point. Callbacks the agent promised land in a queue where one person claims a row, so two people never ring the same customer.

Calls whose audio was too degraded to judge fairly are now flagged and factored into scoring, so an agent is not marked down for a fault that belongs to the phone line. Call detail shows a quality banner, and the flag rolls up per agent, team and phone line — which is how you find the faulty handset or trunk behind it.

Admins can now build multi-step learning paths from real calls, and agents work through them in a dedicated Training area. New agents are enrolled automatically the moment their account is created, badges and completion notifications fire as they progress, and every path shows live enrolment stats — enrolled, completed, and stalled. Managers can see exactly who has stalled and nudge them with a personal message without leaving the admin panel.

A new Staffing Coverage dashboard lays your call activity over a 7-day, 24-hour grid so you can see where volume outran the agents on the floor. Understaffed slots come with a recommended headcount and a "short by" figure, spike weeks are flagged and held out of the baseline, and SLA health renders as a heatmap you can break down per team. All times display in your company's timezone.

Billing teams can connect their Outlook mailbox and have Call Coach IQ draft replies to routine billing email, grounded in the customer's own account record. Every draft queues for human review before anything sends, admins can customise the AI prompt that writes them, and drafts left waiting too long raise an alert. Connection is self-service per company — no IT ticket required.

The platform now continuously scores every agent for burnout risk. When an agent's score reaches Critical, their manager receives an email alert, a "Critical for N days" streak badge appears on the burnout dashboard, and a persistent action panel lets managers log notes and acknowledge the alert to pause repeat notifications. Alert delay and re-alert intervals are configurable per company in admin settings.

Call Coach IQ now detects when more than one agent handles a single call — warm transfers, conference handoffs, team calls — and automatically creates a separate call record for each participant. Every agent gets their own score, AI coaching, and conversation dynamics, so no one's performance is hidden inside a shared call.

A dedicated scoring rubric is now available for warm-transfer calls, with criteria tuned to the handoff context rather than a standard call. Admins can toggle immediate coaching feedback on or off for this call type independently of other rubrics.

Call Coach IQ now connects directly to 19 phone platforms — Intermedia, RingCentral, NICE CXone, Genesys Cloud, Five9, 8x8, Talkdesk, 3CX, Aircall, Dialpad, CloudTalk, JustCall, Freshcaller, MiaRec, Zoom Phone, GoTo Connect, VitalPBX, Twilio, and VoIP.ms. Recordings pull in automatically the moment a call ends. Preferred-tier platforms also stream live agent status and queue statistics into the real-time dashboard. Setup guides and credential forms for every provider are available in Admin → System & Data.

Agents can now submit documents and articles to the Knowledge Base directly from their workspace. Submissions enter a manager approval queue before going live — so agents can contribute their expertise without bypassing editorial control.

When the public visitor chat widget answers a question using your Knowledge Base, it now shows which articles it drew from. Visitors can click through to the source, and you can see exactly which content is driving engagement.

The Call Coach IQ mobile app is now on Android. Cloud builds are fully configured so the team can ship new versions directly to Google Play, and preview builds can be uploaded to Play Console in one step.

JavaScript fixes and improvements are now delivered to the Android app instantly — no app store download required. Users get the latest version the next time they open the app.

The Android app now shows the official Call Coach IQ brand icon and splash screen, replacing the earlier placeholders.

Long-running admin jobs — including embeddings reindex and data backfills — now stream real-time progress so operators can see exactly where a job stands. Jobs can also be cancelled mid-run without waiting for them to finish.

Admins can now see calls per agent, total talk time, average handle time, longest call, calls/hour, and a 30-day score sparkline — all sortable, with stack-rank badges for top and bottom performers.

Click into any agent in the workload tab to see a 7×24 hour-of-day × day-of-week heatmap of their call activity. No extra query — it reuses the same data the table fetches.

A new card surfaces calls that ran more than 2× the agent's median handle time for that call type — so coaches can spot the calls that actually deserve a closer look.

Agents can challenge any AI-generated score directly from the call page. Managers review, respond, and resolve in-thread — every step is tracked.

Built-in gamification layer — agents earn badges for performance milestones, with company-wide celebration alerts.

See a customer's entire call history in one timeline — sentiment shifts, CSAT scores, and churn-risk flags side by side.

Most calls now complete the full transcribe → classify → score → analyze pipeline in minutes.

Natural-language search across your call corpus — ask "show me churn-risk calls from last week" and get a grounded answer with links back to the source calls.

Fully editable scoring rubrics, including separate rubrics per call type (sales, cancellations, tech support, general), are available on Starter and up.

Card numbers, SSNs, PINs, dates of birth, and security answers are now replaced with labeled placeholders before any transcript is stored.