Skip to main content

AI & Analytics

Nexus includes an AI layer that annotates conversations, generates insights, and powers dashboards for customer-service performance monitoring.

AI capabilities​

Conversation Annotation​

The ai-cs-annotator Edge Function analyzes closed conversations and produces structured annotations:

  • Sentiment — positive, neutral, negative.
  • Topic tags — product inquiry, complaint, shipping issue, return request, etc.
  • Risk flags — churn risk, escalation needed, compliance concern.
  • Resolution quality — was the issue resolved? How many messages did it take?

Voice Transcription​

The voice-transcription-worker Edge Function transcribes audio messages (WhatsApp voice notes) into text. Supports Arabic and English. Transcriptions are stored alongside the original message and fed into the annotation pipeline.

CRM Embeddings​

The crm-embeddings-worker generates vector embeddings for customer profiles and interaction history. These embeddings power:

  • Buyer recommendations (crm-recommend-buyers) — suggest contacts likely to purchase a product.
  • Segment discovery — cluster contacts by behavior patterns.

Business Intelligence​

The ai-business-analysis Edge Function generates business analysis reports including sales trends, customer behavior patterns, and operational insights.

Social AI​

The ai-social-generate-report Edge Function analyzes social media performance and generates actionable reports.

Item Intelligence​

The item-intel-worker processes product data through a knowledge graph (Neo4j) to surface product relationships, recommendations, and market positioning.

Analytics dashboards​

CS Dashboard​

The Customer Service Dashboard (/cs-dashboard) shows:

  • Agent response times (average, P95)
  • Conversation volume by channel
  • Resolution rates
  • Customer satisfaction proxies (sentiment distribution)
  • AI-flagged risk conversations

Messages Dashboard​

The Messages Dashboard (/messages-dashboard) shows:

  • Message volume over time by channel
  • Inbound vs. outbound ratio
  • Template message performance
  • Unread/open conversation counts

Call Analytics​

Call analytics show:

  • Call volume and duration trends
  • Missed call rates
  • Agent availability and utilization
  • CDR-based reporting

Shopify KPIs​

E-commerce KPIs synced from Shopify:

  • Order volume and revenue
  • Average order value
  • Fulfillment cycle time

Data flow​

Conversations close → ai-cs-annotator → annotations stored
Voice messages      → voice-transcription-worker → text stored
Customer profiles   → crm-embeddings-worker → embeddings stored
Products            → item-intel-worker → Neo4j graph updated
Annotations + data  → Dashboard queries → CS Dashboard UI

Privacy and safety​

  • AI models process data within the Nexus backend (Supabase Edge Functions).
  • No customer PII is sent to third-party AI services without explicit configuration.
  • All AI outputs (annotations, tags, risk flags) are recommendations — human review is always available.
  • AI features can be disabled per organization via settings.