The AI capability of suggested replies in Deskhero works by analyzing and processing various data sources related to the rehabilitation center for the disabled. This includes previous tickets, knowledge base articles, and even scraped websites for valuable knowledge. The system utilizes OpenAI Embeddings API to generate embeddings vectors for each ticket, article, or website data. These vectors are then stored in a vector database.
Use case 1
When a new ticket is received, the AI system compares the query against the stored vectors to find relevant and similar data. It then uses ChatGPT, an AI language model, to provide an answer based on the attached data. For example, if a ticket asks about the success rate of a specific rehabilitation program, the AI system can suggest a reply that includes statistical data and success stories from previous patients.
Use case 2
In addition to ticket responses, the AI capability of suggested replies also extends to the search bar functionality within Deskhero. When users search for information, the system utilizes traditional search methods as well as OpenAI Embeddings to return the best possible matches. If the search query is formatted as a question, the results can also be used to generate human-like replies based on relevant parts of the matched customer data. This ensures that rehabilitation centers can provide comprehensive and accurate information to their clients.
Use case 3
The AI capability of suggested replies can also be utilized to improve the knowledge base of the rehabilitation center. By analyzing uploaded documents such as policies, guidelines, and manuals, the system can generate relevant articles with rich formatting options. This helps to create a comprehensive knowledge base that can be easily accessed by both support agents and clients.
