Use GenAI to Drive Business Value in Contact Centers

Discover how Slalom and Virgin Voyages are deploying AI-powered chatbots to improve customer service

Photo by Olha Ruskykh via Pexels

Generative AI will bring transformative changes in contact centers, both in the short and long term, but how? At a recent event, Slalom presenter Mark Miller showcased the potential of generative AI (GenAI), demonstrating its use in enhancing business value, notably in Salesforce’s collaboration with Virgin Voyages. Lasting three hours, the event featured live demonstrations, real-world case studies, and insightful discussions focused on practical applications of generative AI in contact centers. Miller shared strategies for optimizing GenAI and addressing challenges, and discussed its transformative impact on contact centers in both short- and long-term scenarios. Find the highlights from our session below, and request your own demo here to learn more.

Meet Vivi, Virgin Voyages’ AI-Powered Assistant

Vivi is a Soul Machines avatar on top of a Salesforce GenAI bot that Slalom built with the help of our Salesforce GenAI framework.

Why Slalom used GenAI for Virgin Voyages: The goal was to leverage Salesforce Knowledge articles to give branded and well-formed responses but use GenAI to understand the customers’ questions. The results we wanted:

• The response is only relevant to the question being asked versus the entire knowledge article being shared with the customer.
• The customer interaction is more conversational.
• The responses are more empathetic as well as on brand.
• Any multipart questions are understood clearly and answered by fetching the knowledge article along with the tone of the brand and fact-checking included by the framework.

Key Challenge & Resolution

Challenge: Slalom faced the task of ensuring that the current knowledge base was prepared, including attaching relevant questions to all knowledge articles.

Resolution: To address this challenge, Slalom developed an extension to our GenAI framework, achieving the following within a few days:

• Generated new, distinctive, and efficient questions for existing articles, promptly incorporating the revised articles back into the review process.
• Scoured chat transcripts to identify question-answer pairs that were not being addressed by the bot and created new articles based on these examples.

This effort would eventually lead to low governance, high accuracy, and knowledge articles that were kept up to date.

Results

A day after the bot was released, there was an event that affected many Virgin Voyages customers as a cruise was canceled. Virgin Voyages published a new article in response to the cancellation giving specific instructions to the customers through the bot. This bot handled 1,200 interactive sessions at a given time. All those interactions did not flow to their contact center, so their call transfer to agents decreased by approximately 250%.

Here are some quotes shared by Virgin Voyages:

“Vivi is just doing SO GOOD. She handles ALL the easy stuff. Sailor Services being very vocal about handling so much less FAQs”

“Noticeable decrease in escalations. Really helping with our headcount discussions for 2024”

“We need to keep pouring gas on this, it’s working just so well”

“So good at recognizing a knowledge gap and making sure we focus on the right new content to curate for knowledge” (edited)

What is Slalom’s GenAI framework?

This powerful tool is designed to equip organizations with the ability to create AI-powered, function-specific bots on the Salesforce platform that speak truthfully and keep their data private. By harnessing the power of the Salesforce Platform and generative AI, this solution aims to solve the challenges of building, validating, and deploying bots for countless scenarios across the enterprise.

Features

The framework provides integration capabilities connecting Salesforce with a large language model (LLM), a vector database, and a chatbot. It reads and summarizes case details, and supports bot creation, fine-tuning, and removal of responses for effective training. The Gen AI bot responds to questions using the LLM and the vector database on web/mobile platforms. Additionally, it provides conversation summaries, chat transcripts, transaction logs, language translation, real-time chat training, and comprehensive reports and dashboards summarizing usage, training, and bot interactions.

Key Components of the Framework

1. Trust: Keeps data private and interactions secure with use of private APIs
2. Guardrails: Ensures responses remain on-brand, on-topic, and on-task
3. Content targeting: Ensures responses are relevant and up-to-date with the latest information about the organization
4. Feedback & refinement: Ensures the bot is receiving a continuous feed
5. LLM agnostic: Connects to your LLM of choice

These components are critical for an effective and responsible generative AI implementation.

Other discussion points:

• Can the chatbot do sentiment analysis? This GenAI chatbot can identify keywords, phrases, and linguistic patterns that indicate different sentiments. This analysis helps the chatbot respond appropriately, understanding the user’s mood or intent and tailoring its responses accordingly. The sentiment can also be plugged into the routing, so the receiving agent is informed.
• Fresh data feed is important: LLMs learn from the data they are trained on. If the data is outdated, the model might make incorrect predictions or provide inaccurate responses because it’s not aware of the latest trends or developments. Fresh data ensures that the model’s predictions and responses remain relevant to current situations and user queries.
• Should the chatbot be used for transactional responses or just informational conversations? The bot can be used for both informational and transactional conversations, but careful consideration must be given to the complexity of the interactions, security requirements, and the accuracy of responses. The goal should be that only the critical and complex queries are referred to the agent while the rest can be handled by the bots.
• What other GenAI use cases is Slalom coming across? One of the critical use cases being discussed is the incident response engine where the bots can leverage the Salesforce Knowledge base containing solutions to a specific critical incident. This can help route the traffic to more self-service versus involving the agents.

Conclusion

Using GenAI for contact centers can open many doors in an organization. Request a demo today to learn about how various industries are utilizing chatbots and GenAI technologies and to see it in action.

Slalom is a global consulting firm that helps people and organizations dream bigger, move faster, and build better tomorrows for all. Learn more and reach out today.