Example: A travel agent using Mystic

A travel agent, let's call her Sarah, is new to the SSP Manager platform. She's trying to understand how to set up a "Shop Category" to filter which airlines' fares she wants to display to her clients. Instead of sifting through lengthy documentation or waiting for a support agent to respond, Sarah opens the chat window and types her question: "What is a shop category?"

Mystic, the chatbot, instantly analyzes Sarah's question using Google's Large Language Models (LLMs) and understands that she's asking about a specific feature of the SSP Manager. It then searches through Mystifly's internal documentation, which has been indexed and stored in a Google Cloud Storage bucket.

Within seconds, Mystic responds with a clear and concise answer:

"In Smart Selling Platform (SSP), Shop Category is the second step of configuration. It brings in the capability to create buckets/categories of inventory from a given supplier, which helps to control and manage inventory sourcing while the flight search is performed. Essentially this configuration determines which supplier's search results should be displayed when a client/agent performs a flight search."

Sarah is satisfied with the answer and proceeds to configure her Shop Category.

Partnering with Google Cloud allows Mystifly to leverage their expertise with Large Language Models (LLMs) and cloud-based solutions for deep content understanding and sophisticated logical reasoning.

Businesses face increasing pressure to achieve more with less, but customer expectations and demands have never been stronger. Today's businesses require a solution to increase productivity, effectiveness, and responsiveness while also tackling operational concerns, such as cost minimization, risk reduction, security, and compliance.

Mystifly's SSP Manager platform is a powerful tool for managing customer onboarding. However, even with detailed documentation available, a first-time user will have questions as airline nomenclature is tricky and can't be understood easily if they are new to the industry. Users might overlook existing solutions or require clarification on specific labels or workflows. This previously resulted in a reliance on support agents, impacting efficiency.

Partnering with Google Cloud for the first time for the two-week GenAI Jumpstart program, Mystifly aims to address these challenges with the development of the Chatbot named ‘Mystic.’ By implementing Google Cloud Vertex AI Agent Builder, Mystic can address customer queries faster.

How ‘Mystic’ works?

Vertex AI Agent Builder is a key component of Mystic, which provides a single powerful platform for building search, conversational and generative AI solutions on Vertex AI. This combination of technologies allows the creation of virtual agents that can understand and respond to natural language in a human-like way, even for complex requests. But before that, the following prerequisites need to be satisfied:

• The internal Mystifly PDFs are uploaded in the Google Cloud Storage bucket.
• An unstructured datastore is created to intelligently index data within these internal PDFs to make relevant information easily accessible to the end user.
• Finally, the chat app is linked to this datastore to understand and respond to user queries in a natural and intuitive way.

Following is the overall solution workflow in points:

• User asks a question: The user interacts with the search engine or chatbot, asking a question about SSP Manager configuration.
• AI processes the query: Google's AI technologies analyze the question, understanding its intent and context.
• Relevant information is retrieved: Foundational LLMs extract the most relevant information from the unstructured data store, pinpointing the answers contained within PDFs.
• Summarized response is provided: Then, a clear, concise summary of the answer, along with links to the citation document pages are presented for further reference. 
• Chatbot escalation (if needed): If the user still requires assistance, the chatbot provides an option to trigger an email via Dialogflow CX Webhook and Cloud Functions, thereby, connecting the user with a support agent.

Benefits of the Google Cloud partnership

• Enhanced user experience: SSP Manager users can now resolve most of their configuration queries independently, leading to increased satisfaction and reduced frustration.
• Improved efficiency: Support agents are freed from answering routine questions, allowing them to focus on more complex issues.
• Scalability: Google Cloud's AI solutions can easily scale to accommodate Mystifly's growing customer base and evolving needs.

Example conversation

Chatbot example screenshots are as follows:

User: What is the shop category?

AI: In Smart Selling Platform (SSP), Shop Category is the second step of configuration. It brings in the capability to create buckets/categories of inventory from a given supplier, which helps to control and manage inventory sourcing while the flight search is performed. Essentially this configuration determines which supplier's search results should be displayed when a client/agent performs a flight search.

Example: Mail webhook trigger

We trigger a webhook in case the chatbot is not able to answer the user query sufficiently.

AI: I apologize, I cannot answer your question at this time. Perhaps you could try asking in a different way, or connect with our support team at ​​support@mystifly.com.

When you click on the mail ID, it will result in the following pop-up:

When you click on ‘Open Mail,’ it results in the following placeholder where the email can be typed by the end user.