Customers move between email, chat, phone and social media expecting continuity. In most companies, each channel is a separate silo — the customer repeats their story at every handoff, and agents lack context from prior interactions.

Utilyx models omnichannel support as an executable workflow: a 24/7 AI chatbot auto-resolves from the KB, escalations to agents carry full context, and email, chat, phone and social feed a single unified queue with a shared customer timeline.

The workflow is structured as a decision tree with 3 branches (AI auto-resolve, agent escalation, supervisor review) and 8 steps end-to-end. Every conversation is logged in the chat_audit subtable on the customer record. Customers repeat their story 3+ times across channels — unified context stops this.

• Step 1 (Start — Intake): conversations arrive from email, chat, phone (transcribed via call-connector), social media and in-app messaging into a single unified queue with mandatory fields conversation_id, customer_id, channel, message, language, sentiment_score. All channels feed the same customer_timeline.

• Step 2 (AI chatbot): the copilote IA handles conversations 24/7, answering from the knowledge base with semantic search (vector embeddings via sql-query against kb_article_index) across 7 languages. It resolves common queries without an agent and scores confidence (0-100).

• Step 3 (Branch — confidence split): Branch A (confidence > 85): auto-resolve — the chatbot answers, the case is closed and the deflection is logged. Branch B (confidence 50-85): suggest answer + offer agent transfer. Branch C (confidence < 50 or negative sentiment): immediate escalation to agent with full context.

• Step 4 (Auto-resolution): when the chatbot resolves the query, the case is closed and deflection_event is logged (conversation_id, customer_id, channel, article_id); the conversation is saved to the customer_timeline.

• Step 5 (Escalation with context): when the chatbot cannot resolve, it escalates to an agent with the full conversation history, customer profile, sentiment analysis, suggested KB articles and prior ticket history — no repeated questions. The unified agent desktop shows every channel in one view.

• Step 6 (Routing): visual routing rules assign conversations by skill, language and priority via sql-query against agent_skill_matrix. The AI copilot builds the routing logic from a plain-language description.

• Step 7 (Supervisor review — sub-branch): if sentiment_score < -0.5 or handle_time > 10 min, the conversation is auto-flagged to a supervisor for quality review, logged in supervisor_review_queue.

• Step 8 (Closure & feedback): after resolution, a CSAT survey is auto-triggered on the same channel. The complete conversation and CSAT are archived in useArchive with index fields conversation_id, customer_id, channel, date for compliance.

Teams using Utilyx deflect up to 40% of conversations via the 24/7 AI chatbot and cut average handle time by 30% thanks to full context on escalation — no repeated customer story across the 3+ channels a typical customer touches.

The no-code visual designer lets a customer service manager build and adjust routing without developers, and the unified queue eliminates channel silos — 60% faster deployment at 70% lower TCO than legacy omnichannel suites. The confidence-based branching ensures the chatbot only auto-resolves when it is genuinely confident.

The shared customer_timeline means every agent, on every channel, sees the same conversation history — ending the 3x story repetition that frustrates customers. 7-language support lets one workflow serve a multilingual customer base without separate deployments.

Semantic search across the KB ensures the chatbot and agents surface the right answer, not just keyword matches, and the supervisor review queue catches quality issues before they damage CSAT.