Skip to main content

Customer Service Knowledge Base ChatBot

Starcom AI Knowledge Assistant Screenshot

Company: Starcom - Software for plant nurseries, for over 30 years.

Challenge: Starcom's three decades of success had resulted in a very sophisticated and full-featured piece of software, including customer-specific features. There was plenty of documentation, but it was often difficult to know what to search for and where. Knowledge was spread across SharePoint Wikis, project management systems, and even recorded meetings—making the CEO the go-to resource for complex or nuanced questions, even requiring their attention for Customer Support inquiries. This created a bottleneck where the CEO's time was increasingly spent on knowledge-sharing rather than strategic initiatives. Additionally, onboarding new Customer Support agents was requiring months of training, even limiting growth opportunities.

Approach: We created an easy-to-use AI Chatbot and provisioned it with all the available information from across their knowledge ecosystem:

  • Their SitePoint Wiki
  • Development issue tracker
  • Customer support issue tracker
  • PowerPoint presentations
  • Markdown and Word docs from code repositories
  • Transcribed video meetings (yes, we converted all those recorded calls!)
  • Archived systems

Some documentation was outdated, but the manhours required to cull it were dauting, so we levergaged an LLM to sift through it. The resulting AI doesn't just search the content, it understands the context, connected related information, and can answer questions the way someone with institutional knowledge (like the CEO) would.

Results:

  • CEO freed to focus on strategic leadership and business development
  • Faster Customer Service Member onboarding with the AI acting as a "backstop" level of understanding.
  • Faster Answers: All team members gained instant access to previously burried knowledge
  • Increased employee independence and confidence
  • Expedited foundational knowledge learning for a new Software Engineering project.

Technologies Used:

  • OpenAI GPT-4o
  • Retrieval Augmented Generation (RAG)
  • Embedded web-based AI
  • Multi-format document ingestion pipeline
  • Automated Video Transcription,
  • MSSQL Database on Azure
  • SharePoint
  • And, some custom code to glue it all together.