Generative AI for Customer Service Training Course

Participants can map AI-assisted journeys with clear responsibilities, decision points and recovery paths.

• Service blueprint mapping across customer actions, channels and backstage processes
• Moments of truth identified through intent, emotion and customer effort
• BPMN 2.0 models for routing, exceptions and escalation decisions
• Journey orchestration triggers from customer intent, profile and interaction history
• Designing channel transitions between chat, voice, email and human support
• Recovery paths for failed automation, missing information and dissatisfied customers
• Ownership matrices defining service, technology, risk and content responsibilities

Participants can specify a governed knowledge foundation for grounded AI customer responses.

• Retrieval-augmented generation architecture for policy-grounded customer responses
• Knowledge-source inventory covering policies, procedures, products and approved scripts
• Content chunking methods using headings, semantic boundaries and metadata
• Metadata taxonomy for product, jurisdiction, customer segment and content ownership
• Retrieval filters enforcing entitlement, language, region and effective-date rules
• Citation patterns showing agents and customers the response evidence
• Knowledge lifecycle controls for versioning, review, retirement and audit trails

Participants can create prompt patterns and workflow steps for reliable service tasks.

• Prompt anatomy: role, task, context, constraints, format and examples
• Few-shot prompting for classification, tone adaptation and response drafting
• Structured outputs using JSON schemas for downstream service workflows
• Chain-of-verification prompts for checking claims against retrieved evidence
• Function calling patterns for CRM lookup, case creation and status updates
• Prompt libraries governed by version, owner, test evidence and approval
• Instruction hierarchy handling customer input, business rules and safety constraints

Participants can design appropriate AI assistance levels and controlled human escalation routes.

• Agent-assist patterns for summarisation, suggested replies and after-call work
• Self-service boundaries defined by decision impact, confidence and customer vulnerability
• Human-in-the-loop checkpoints for approvals, exceptions and sensitive outcomes
• Escalation rules using confidence thresholds, intent risk and repeat contacts
• Agent interface requirements for citations, edits, overrides and feedback
• Conversation memory policies balancing continuity, relevance and privacy obligations
• Queue-routing logic connecting AI classification with skills-based assignment

Participants can apply generative AI to sales support while preserving approval and customer-contact controls.

• Sales journey stages from lead qualification through renewal and expansion
• AI-assisted account research using approved internal and public information
• Lead scoring signals separated from generated narrative and recommendations
• Next-best-action prompts aligned with sales playbooks and customer consent
• Drafting follow-up messages with brand, regulatory and offer constraints
• Opportunity summaries combining CRM records, meeting notes and service history
• Approval workflows for pricing, commitments, regulated claims and outbound contact

Participants can define governance controls for customer-facing generative AI services.

• NIST AI RMF functions applied to customer-facing generative AI services
• ISO/IEC 42001 controls for accountability, documentation and continual improvement
• Personal data minimisation across prompts, retrieval stores and conversation logs
• Consent, disclosure and identity-verification requirements by interaction type
• Bias testing for language, accessibility, vulnerability and customer treatment
• Safety policies addressing harmful content, manipulation and unauthorised advice
• Incident management procedures for unsafe outputs, data exposure and service failures

Participants can test AI service performance using structured datasets, metrics and review procedures.

• Evaluation datasets sampled from intents, edge cases and historical failures
• Groundedness scoring against retrieved evidence and approved policy content
• RAGAS measures for faithfulness, answer relevance and context precision
• Service quality rubrics covering accuracy, tone, resolution and effort
• Red-team testing for prompt injection, jailbreaks and data extraction
• A/B test design comparing AI journeys with existing service processes
• Quality assurance sampling combining automated checks and expert review

Participants can define operating responsibilities and integration requirements for an AI service deployment.

• Target operating model for product ownership, service operations and model oversight
• RACI assignments across business sponsors, agents, engineers and risk reviewers
• Vendor due diligence covering data handling, model changes and service levels
• Integration patterns linking CRM, knowledge bases, telephony and case management
• Observability requirements for prompts, retrievals, outputs, overrides and incidents
• Change-management process for policy updates, prompt releases and model revisions
• Service-level metrics covering latency, availability, containment and escalation rates

Participants can prepare a phased implementation plan for a governed AI service pilot.

• Pilot scope defined by journey, customer segment, channel and decision boundaries
• Minimum viable knowledge set with accountable owners and review dates
• Release criteria combining quality thresholds, risk controls and operational readiness
• Phased rollout plan using limited exposure, monitoring and controlled expansion
• Benefits tracking baselines for handling time, resolution and customer effort
• Training plan for agents, supervisors, content owners and escalation teams
• Ninety-day action plan assigning decisions, dependencies and review milestones