AI-Powered Workflow Automation for Customer Support Teams

Role: You are an AI workflow automation specialist for a customer support department.

Objective: Design a comprehensive, AI-powered workflow for handling a specific, frequently occurring customer support inquiry within a B2B SaaS environment. The goal is to maximize resolution efficiency, reduce human agent workload, and maintain high customer satisfaction.

Required Inputs:
1. [Specific Inquiry Type]: Clearly define the exact customer inquiry this workflow will address (e.g., 'Password Reset Requests', 'Billing Inquiry for Overdue Invoices', 'Feature Availability Check for Basic Plan').
2. [Current Manual Workflow Steps]: List the existing step-by-step process a human agent currently follows to resolve this inquiry, including any tools used, information gathered, and decision points.
3. [Key Customer Data Points]: Identify essential pieces of customer information typically required to resolve this inquiry (e.g., 'Account ID', 'Email Address', 'Subscription Plan', 'Last 4 digits of Credit Card').
4. [Resolution Criteria]: Define what constitutes a successful resolution for this inquiry (e.g., 'Password successfully reset and confirmed', 'Invoice explanation provided and payment link sent', 'Feature status clearly communicated').
5. [Escalation Triggers]: Specify conditions under which the AI agent should escalate the inquiry to a human agent (e.g., 'Customer expresses frustration multiple times', 'AI unable to find required customer data', 'Inquiry falls outside predefined scope').
6. [Company Tone and Brand Guidelines]: Describe the desired communication style and any specific phrases or policies the AI agent must adhere to.
7. [Available Integrations/Tools]: List any internal systems the AI agent could potentially access or interact with (e.g., 'CRM database', 'Knowledge Base', 'Billing System API', 'Support Ticket System').

Execution Steps:
1. **Analyze Current Workflow:** Deconstruct the provided manual workflow steps, identifying bottlenecks, repetitive tasks, and decision points.
2. **Design AI Interaction Flow:** Develop a step-by-step conversational flow for the AI agent, from initial greeting to resolution or escalation. This flow should incorporate data gathering, information retrieval, problem-solving logic, and dynamic responses.
3. **Define Data Requirements & Access:** Specify exactly which customer data points the AI needs at each stage and how it would access or request them. Detail any necessary system integrations.
4. **Craft AI Responses & Prompts:** Write example AI responses and internal prompts for key stages, ensuring they align with the specified tone and brand guidelines. Include prompts for clarifying customer intent and confirming information.
5. **Establish Escalation Protocol:** Detail the exact conditions and procedures for escalating to a human agent, including what information should be automatically packaged and transferred to the human agent.
6. **Outline Guardrails & Policy Adherence:** Describe how the AI will ensure compliance with company policies and legal requirements, especially regarding sensitive customer data.
7. **Suggest Performance Metrics:** Recommend key performance indicators (KPIs) to measure the effectiveness of the AI-powered workflow (e.g., 'Resolution Rate', 'Average Handling Time', 'Customer Satisfaction Score').

Output Contract:
Provide a detailed, structured document outlining the AI-powered workflow. The output should include:
1. **Workflow Title & Description:** A concise overview.
2. **AI Interaction Script:** A full, branching conversational script detailing AI prompts, expected customer responses, and subsequent AI actions/responses. Use clear indentation for decision trees.
3. **Integration Points:** A list of required system integrations and the specific actions the AI would perform within each system.
4. **Escalation Procedure:** A clear protocol for human agent handoff.
5. **Guardrails & Compliance Notes:** Specific instructions for maintaining policy adherence.
6. **Recommended KPIs:** A list of metrics for monitoring performance.
7. **Potential Challenges & Mitigation:** Identify anticipated issues and suggest solutions.

Constraints:
- The workflow must prioritize customer data security and privacy.
- All AI responses must be clear, concise, and empathetic.
- The solution should aim for at least 70% automated resolution for the specified inquiry type.
- Assume access to a robust, configurable AI agent platform capable of natural language understanding, database lookups, and API calls.

Quality Check: Does the proposed workflow address the [Specific Inquiry Type] comprehensively? Is every step logical and clearly defined? Does it effectively minimize human intervention while ensuring a positive customer experience and adherence to company standards? Is the escalation path unambiguous?