Why it's useful:
- Provides instant answers to customers 24/7 without human intervention or phone queue delays
- Deflects routine enquiries, freeing your service team to focus on complex issues
- Automatically identifies knowledge gaps by flagging unanswered questions with prompts to create new articles
- Enables personalised service by accessing CRM data (order history, customer preferences, account details)
- Scales your service capacity without proportionally scaling headcount
- Continuously improves as you fill knowledge gaps based on actual customer questions
The challenge: Service teams face mounting pressure to respond faster while maintaining quality. Customers expect immediate answers, but human agents can't be available 24/7 for every query. Meanwhile, knowledge bases often contain gaps that only become apparent when customers ask questions you can't answer. Manual identification of these gaps is reactive and time-consuming. Teams need a solution that both deflects routine queries AND systematically improves service resources.
How to implement it:
Set up the Customer Agent foundation:
- Navigate to Service > Customer Agent in HubSpot
- Give your agent a name and avatar that aligns with your brand personality
- Define the agent's tone and personality (professional, friendly, technical, etc.)
- Set the agent's primary language and regional settings
Train the agent with comprehensive knowledge:
- Connect your HubSpot knowledge base articles as the primary information source
- Add your blog posts, landing pages, and website content
- Upload support documentation, product guides, and FAQs as PDF files
- Include external public URLs (company website, help center, product pages)
- Key principle: An AI agent is only as good as the data it's trained on - be thorough
Integrate CRM data for personalisation:
- Enable access to specific CRM properties (order history, subscription tier, account status, preferences)
- Configure which data points the agent can view and share with customers
- For sensitive information (payment details, personal data), set up verification requirements or automatic handoff to human agents
- Test data access to ensure privacy compliance and appropriate boundaries
Create Actions for system integration:
- Use the Actions feature to connect external systems via API
- Configure common use cases: package tracking, order status checks, appointment booking, account balance queries
- Set up data push/pull workflows so the agent can retrieve real-time information
- Test each Action thoroughly before making it available to customers
Configure intelligent handoff triggers:
- Build keyword triggers that escalate to human agents (e.g., "angry," "frustrated," "speak to a person," "complaint")
- Set up emotion detection for sentiment-based handoffs
- Create complexity thresholds - if the agent can't resolve after X exchanges, handoff automatically
- Define handoff routing rules to send conversations to the right team or specialist
Implement knowledge gap resolution workflow:
- When the agent flags "Unable to answer," review the customer's question
- Use HubSpot's built-in prompt to either:
- Create a new knowledge base article for complex topics
- Add a "Quick Answer" for simple, frequently-asked questions
- Assign content creation to the appropriate subject matter expert
- Monitor the "Unanswered Questions" report weekly to prioritise gap-filling
Test and deploy strategically:
- Use the test function extensively before going live (note: Customer Agent uses credits)
- Start with a pilot deployment on lower-stakes channels (e.g., specific website pages)
- Monitor key metrics: deflection rate, conversation handling success, handoff frequency, customer satisfaction scores
- Gather feedback from both customers and the human agents receiving handoffs
Optimise continuously:
- Review conversation transcripts monthly to identify patterns in questions the agent struggles with
- Refine agent instructions based on real usage patterns
- Update knowledge sources as products, policies, or services change
- Track which knowledge gaps get resolved fastest and apply those lessons to future content creation
Result:
- Service teams deflect 40-60% of routine enquiries automatically, freeing capacity for complex issues
- Customers receive immediate answers 24/7, improving satisfaction and reducing wait times
- Knowledge base improves systematically based on actual customer questions, not guesswork
- Service quality becomes more consistent as the agent delivers standardised, accurate information
- Team can handle higher query volumes without additional headcount
- Data-driven insights reveal which topics generate the most questions, informing product documentation and training priorities
