Case study
An AI Sales Assistant Grounded in Product Knowledge
Product answers and a support handoff in one conversation.
The challengeHelping learners navigate product and sales questions.
The solutionKnowledge-grounded answers with ticket-based escalation.
Product and internal knowledge connected to website chat.
HubSpot support-ticket creation for unresolved requests.
Approximately 31,571 messages per month reported.
Engagement
Overview
McKissock Learning needed a customer-facing AI assistant to help prospective learners navigate product and sales questions. The goal was to make relevant company knowledge available through conversation while keeping a route to human support.
I built an agentic retrieval-augmented generation workflow that connects knowledge lookup with tool-based support escalation. The assistant retrieves context from website content, product information, and internal documents to support its answers. When a request remains unresolved, it can create a HubSpot support ticket.
The system combines retrieval-grounded answers, support-ticket escalation, and request logging. New Relic provides visibility beyond the chat interface, while the ticket-creation tool connects unresolved conversations to the support workflow.
Usage figures are internally reported. Reporting periods and user definitions remain unconfirmed.
- Reported users
- 4,085
- Reported messages / month
- ≈31,571
Challenge
The assistant needed relevant business context and a practical next step when a question required more help.
Company-specific answers
Product and sales questions needed context from website content, product information, and internal documents.
Unresolved requests
The workflow needed to connect learners with support rather than stop at a conversational response.
Operational visibility
Requests needed logging so the interaction could be observed outside the chat interface.
Solution
I connected knowledge retrieval, a support action, and request logging in one customer-facing workflow.
Knowledge retrieval
Look up relevant company material to support the assistant’s responses.
Conversational delivery
Make the assistant available through the McKissock website chat experience.
Tool-based escalation
Invoke a dedicated ticket-creation tool for unresolved requests.
Support integration
Create tickets in HubSpot to continue the request through the support process.
Request logging
Send requests to New Relic for operational visibility.
Section: Agentic RAG with a practical support action
Retrieval-augmented generation connects the assistant’s answers to company knowledge. The agentic element is its ability to use a tool as part of the conversation: when further assistance is needed, it can create a support ticket. This gives the assistant a role beyond answering questions without implying that it handles every request autonomously.
Impact
Reported usage shows recurring adoption in the learner sales journey, with knowledge retrieval and support escalation in the same experience.
Reported adoption
4,085 users.
Reported activity
Approximately 31,571 messages per month.
Implemented capability
Company-grounded responses, HubSpot ticket handoff, and request logging.