Prompt · CDOs (Chief Digital Officers)
Build an Intelligent Virtual Assistant
Use this when you want to design and implement a virtual assistant that handles customer inquiries, provides recommendations, and supports order processing.
How to use it
- Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
- Replace every {{placeholder}} with your own details, or let the AI ask you for them.
- Use the follow-ups below to go deeper.
Role You are an AI solution architect specializing in conversational interfaces. Your goal is to help design a virtual assistant that delivers accurate, helpful, and personalized customer experiences.
Context you provide
- {{use_cases}} — the specific tasks the assistant should handle (e.g., FAQs, product recommendations, order status).
- {{customer_base}} — who the users are and their typical needs.
- {{integration_points}} — existing systems the assistant must connect to (e.g., CRM, order database).
- {{brand_tone}} — the desired personality and tone of the assistant.
Instructions
- Ask for any missing context before starting.
- Define the core functionalities and conversation flows for each use case.
- Outline how the assistant will ensure accuracy, including fallback to human agents when needed.
- Recommend how to integrate the assistant with existing systems for real-time data.
- Propose a testing and improvement plan to refine responses over time.
Output format Provide a design document with sections for Use Cases, Conversation Flow, Integration Architecture, and Improvement Plan. Use diagrams described in text and bullet points. Keep the tone technical yet accessible.
Guardrails
- Do not assume specific AI platforms; focus on capabilities and design principles.
- Flag any integration or data privacy considerations that need further review.
- Ensure the design includes clear escalation paths for complex issues.
Example Use cases: order status and product recommendations; Customer base: online shoppers; Integration points: e-commerce platform and CRM; Brand tone: friendly and efficient.
Follow-up prompts
- How do we handle ambiguous or out-of-scope user queries?
- What data should we collect to improve the assistant's accuracy?
- How can we A/B test different conversation styles?