Move cursor | Click to ripple
Platform anatomy

Inside the platform that scores 3 billion conversations a year.

QEval® is one system: a data layer that ingests every channel, an expert-routed AI engine that runs 326 million classifications every five minutes, and a coaching layer that converts scores into agent behavior change. This page maps the system underneath the score.

11
Modules
One unified data layer, one scorecard
80+
Pre-built integrations
CCaaS, CRM, LMS, WFM, HCM, data, identity
35+
Languages
Regional dialect adaptation
6
Channels
Voice, chat, email, SMS, social, video
The pipeline

Six stages, one data layer.

Every conversation moves through the same pipeline. Ingest from any channel. Redact before any model touches the data. Route to the right expert. Score, coach, and attribute outcome. The same path for a human agent on Genesys and an AI agent on Sierra.

STAGE 01

Ingest

Universal Connector pulls voice, chat, email, SMS, social, and video from 80+ pre-built integrations. No middleware. No ETL project.

80+ sources · 6 channels
STAGE 02

Redact

Named Entity Recognition removes PII, PHI, and PCI at ingest. Nothing sensitive reaches the model. HIPAA, PCI-DSS, GDPR-aligned by default.

Pre-LLM redaction
STAGE 03

Classify

A hybrid MoE + ML classifier routes each scorecard item to the right method. Traditional ML handles high-volume signals like topic and intent. Purpose-trained MoE experts handle nuance: empathy, brand voice, compliance language.

Hybrid MoE + ML · 326M / 5 min
STAGE 04

Score & Generate

Expert sub-models to score compliance, empathy, resolution, and brand voice in parallel. The same pass generates call summaries, case notes, next-best-actions, and training allocation. The agent does not write the wrap-up.

94%+ accuracy · auto wrap-up
STAGE 05

Coach

HI Model converts scores into agent-specific coaching plans tied to the exact transcript moment. Pushed to LMS, supervisor in queue, or directly to the agent.

Delivered within 48 hrs
STAGE 06

Improve

Outcome Attribution traces every coaching intervention to CSAT, AHT, revenue, and compliance metrics. The board-ready proof loop.

Six Layers of Intelligence
326M
Classifications
Every 5 min
94%+
Accuracy
Contractual SLA
98%+
Compliance precision
95%+ recall on flagged
0
3rd-party model exposure
Closed-source MoE
Channel coverage

Every channel you operate. One scorecard.

Customers talk to you on the channel they prefer. The same conversation can start as chat, escalating to voice, and closing on email. QEval® ingests, transcribes if needed, redacts, and scores all six on the same model.

Voice

Real-time and post-call. Diarization, transcription, and sentiment.

Chat

Live web chat, in-app messaging, bot handoffs. Turn-by-turn scoring.

Email

Inbound case threads, outbound resolution, signature compliance.

SMS

Two-way SMS, MMS, asynchronous resolution flows.

Social

Public mentions, DMs, and brand sentiment monitoring.

Video

Vision Model scores screen-share workflows and on-camera tone.

Language coverage
35+
Languages, with dialect adaptation

The Vocabulary Library is tuned to contact center language by region. Spanish for Mexico City is not the same product as Spanish for Madrid. We trained the experts that way.

English variants
USUKAustralianNew ZealandIndian
Western Europe
Spanish (Standard)Spanish (LATAM)French (Standard)French (Canadian)German (Standard)German (Swiss)ItalianPortuguese (BR)Portuguese (PT)Dutch (incl. Flemish)Greek
Nordic
SwedishNorwegianDanishFinnish
Eastern Europe
RussianPolishCzechSlovakHungarianRomanianBulgarianUkrainian
Baltic
EstonianLatvianLithuanian
Asia
Japanese (Tokyo)Japanese (Standard Voice)Korean (Standard)Mandarin+ regional dialects on request
The anatomy

Eleven modules. Three tiers. One platform.

QEval® is not a suite of point products glued together. The same data layer that ingests a Genesys call is what feeds the AI Coach, the gamification leaderboard, and the executive analytics dashboard. Here is how the modules stack.

TIER 01FoundationCapture, redact, store, route
Universal Connector
80+ pre-built integrations. Platform-agnostic ingest.
Redaction Pipeline
PII/PHI/PCI removed at ingest via NER. Pre-LLM.
MoE Engine
Proprietary closed-source mixture-of-experts model.
Classification Engine
Routes each scorecard item to the right expert path.
Vocabulary Library
Industry-tuned lexicons across 10+ verticals, 35+ languages.
TIER 02IntelligenceScore, surface, alert
Auto QA
Contextual scoring of 30 to 100% of interactions.
AI Agent QA
Vendor-neutral scoring across Sierra, Decagon, Agentforce, Ada, and more.
Real-Time Agent Assist
In-call guidance, compliance prompts, knowledge surfacing.
Speech Analytics
Sentiment trajectory, intent detection, and topic clustering.
Vision Model
Auto screen-capture analysis for desktop workflows and video.
TIER 03OutcomeCoach, engage, report, attribute
HI Coaching Lifecycle
Auto-generated coaching tied to interaction moments.
AI Coach
Conversational coach that walks an agent through their gaps.
Gamification
Leaderboards, badges, skill trees. 60% higher engagement.
Surveys + CSAT
Predictive CSAT at 85%+ accuracy across channels.
Analytics + BI
110+ reports. Natural language queries. Six-Layer dashboard.
Coming soon
Video Coaching
Async video coaching tied to interaction moments. Manager-recorded clips delivered into the agent's learning queue.
Coming soon
Intelligent WFM
Forecasting, scheduling, and adherence informed by QEval® performance and quality signals, not just queue volume.
Inside the product

One platform. Six lenses.

The agent sees a real-time coach. The QA analyst sees a calibration workbench. The CFO sees a finance-ready ROI dashboard. Every role gets a different view on the same underlying data.

Agent

Real-time coach in the agent desktop.

Live guidance during the call, plus a personal scorecard, recent coaching feed, and a learning queue.

  • Compliance prompts surfaced in the moment
  • Side-by-side personal scorecard
  • Skill tree and gamification streak
Supervisor

Team scorecard, coaching queue, anomalies.

Watch the team in real time. Triage who needs coaching first, why, and what the exact transcript moment was.

  • Live floor view across all agents
  • Auto-generated 1:1 agendas
  • Compliance and escalation alerts
QA Analyst

Calibration workbench and sample selection.

Side-by-side human vs AI review, scorecard authoring, bias detection, calibration sessions across the team.

  • Sample selection by risk score
  • Disagreement view between human and AI
  • Custom scorecard builder, no-code
Coach + Trainer

Coaching plans tied to interaction moments.

Pull up an agent and see exactly which moments their week was made or broken. Push targeted learning into the LMS.

  • Skill gap heat map by team
  • Linked learning content per gap
  • Outcome tracking on prior coaching
Operations

Volume, AHT, compliance pass rate, capacity.

Cross-channel operational view. Spot the shift, the queue, or the LOB that is drifting and act on it the same day.

  • Channel and queue volume trends
  • Compliance pass rate by LOB
  • Capacity forecasting inputs
Executive

Six-Layer dashboard, finance-ready ROI.

The board view. Quality, customer, revenue, ops, training, strategy intelligence in one place. Exportable for the next QBR.

  • Six Layers of Intelligence rollup
  • Outcome attribution to dollar value
  • Board-ready exports
Integration posture

80+ pre-built integrations. Platform-agnostic. No middleware.

The average enterprise runs 3.9 platforms across telephony, CRM, and workforce management. QEval® ingests all of them in parallel. Add a new platform without rebuilding your scoring program.

80+
Pre-built connectors
9
Categories
10
AI agent platforms scored
0
Middleware required

Telephony / CCaaS

12
Genesys, NICE CXone, Five9, Amazon Connect, Talkdesk, and more

CRM

8
Salesforce, HubSpot, Dynamics 365, Zendesk, ServiceNow, and more

AI Agent platforms

10
Sierra, Decagon, Agentforce, Ada, NICE Enlighten, and more

LMS / Training

7
Cornerstone, Litmos, Docebo, TalentLMS, and more

WFM

8
Verint, NICE WFM, Calabrio, Alvaria, Assembled, and more

HR / HCM

7
Workday, ADP, SAP SuccessFactors, UKG, and more

Data Warehouse

6
Snowflake, Databricks, BigQuery, Redshift, Synapse

Identity / SSO

7
Okta, Microsoft Entra, Auth0, Google Workspace, SAML 2.0

Collaboration + Ticketing

7
Slack, Microsoft Teams, Jira, Zendesk Support, and more
Deployment model

Run QEval® on AWS or Azure. GCP soon.

Multi-cloud, regional data residency, and a single-tenant option for regulated buyers. Pick the cloud your security team already trusts.

Cloud platforms
Amazon Web Services
Multi-region, VPC peering, KMS
Live
Microsoft Azure
Azure regions, Private Link, Key Vault
Live
Google Cloud
In active build, target 2026
Coming soon
Data residency & tenancy
REGION 01
Americas

US East, US West, Canada Central. SOC 2 Type II, PCI DSS, HIPAA-ready.

REGION 02
EMEA

Ireland, Frankfurt, London. GDPR-compliant. Data stays in-region.

REGION 03
APAC

Mumbai, Singapore. India and Singapore residency available.

REGION 04
Single-tenant

Dedicated instance for regulated workloads. VPC peering, customer-managed keys.

Deployment commitment

30 days to live. Fully managed.

No IT dependency. Two dedicated resources assigned to your deployment: a Customer Experience Manager and a Dedicated Analyst. With contractual 60-day exit rights if we do not meet our accuracy of SLAs.

Week 1
Integrate

API connectors to your telephony, CRM, and data systems. Pre-built for Genesys, NICE, Salesforce, and 80+ others. No middleware. No custom code.

Week 2
Configure

Mirror your evaluation criteria. Custom scorecards calibrated against your best human evaluators. Matches your existing evaluation framework — whether 47-point or custom.

Week 3
Validate

Parallel run: AI evaluations alongside human QA. Accuracy benchmarked at 94%+ before go-live. Side-by-side comparisons on your real interactions.

Week 4
Go live

Full deployment. Coaching workflows are activated. Real-time dashboards online. Your team is trained, and the platform starts delivering value from day one.

Post-deployment

We deploy outcomes, not just software.

A 30-day structured utilization program after go-live. We address the six most common adoption barriers and track measurable success criteria until your team is self-sufficient.

Week 1
35%
Week 2
54%
Week 3
67%
Week 4
77%
Week 5
84%
300%
Coaching frequency increase
Measured across enterprise deployments
40%
Faster improvement cycles
Reduced time to measurable agent performance gains.
90%
Supervisor adoption target
Conducting QEval®-referenced coaching calls
APIs + extensibility

Built to be extended, not just used.

Every module is API-first. Pull scores into your BI stack. Push new scorecard items from a CSV. Trigger your downstream workflow off a compliance violation event. Build your own coaching surface on top of QEval® data.

REST
Scoring API
Read scores, scorecards, agent metadata. Write custom scorecard configurations. Versioned, rate-limited, audit-logged.
GET /v1/scores?agent_id=...
EVENT
Webhooks
Subscribe to events. Compliance violation, sentiment crash, coaching delivered, AI agent drift. Fire to your endpoint.
on: compliance.violation
NO-CODE
Scorecard Builder
Author scorecards without engineering. Calibrate AI scoring against your best evaluators. Push live without a release.
+ new criterion
ASR
ASR Vocabulary
Customize the speech recognition vocabulary for your industry jargon, product names, and scorecard terms. Higher recognition accuracy on the words your scoring actually depends on.
vocabulary: tenant-specific

Enterprise security, zero exceptions.

Proprietary closed-source MoE means customer data never enters a third-party foundation model. PII is redacted at ingest, before the model runs. Audit trails on every classification decision.

SOC 2 Type IIISO 27001ISO 42001PCI DSS Level 1HIPAAGDPRCCPA
Run it on your stack

Bring your scorecards. Bring your CCaaS. Bring an AI agent.

We will stand up the platform against your real interaction data and show you what your current program is missing. 30 days to a working deployment. 90-day pilot with 60-day exit rights.

Contractual commitments

Four numbers no peer publishes.

94%+
Accuracy SLA
Written into the master agreement
30 days
Deployment
Money-back guarantee
60 days
Exit clause
Cancel with notice, no penalty
120 days
ROI Window
Value case before expansion