by Carl Bender, Senior Technical Staff Member for Power Hardware Technical Support and Lead Architect for Client Experience AI solutions
Generative AI introduces new capabilities that transform human interaction with technology and with each other, creating more natural, conversational experiences. It allows us to develop chatbots and virtual assistants that help us find information and complete tasks through interactive dialogue, enabling more seamless, human-like communication between people and machines.
What is the Virtual Assistant?
The support community has a long tradition of building the knowledge base for known issues and commonly asked questions which are used by support agents to provide solutions for clients in support cases quickly.
Much of this knowledge is publicly available and searchable on the internet. However, this often does not translate into easy-to-find exactly the right answer for a specific question. Furthermore, our community employs a variety of productivity tools to help support agents debug problems on behalf of clients and assist them with multiple tasks.
By leveraging a comprehensive knowledge base and past interactions, IBM's virtual assistant can efficiently address many inquiries through self-service. If needed, it can seamlessly transition to a live agent or open a ticket for resolution by a support expert, driving faster resolution with less effort.
How-to Questions
Clients don’t want just a better search engine from IBM; they want personalized answers to their questions that only our Subject Matter Experts (SMEs) can provide. We brought together SMEs from each product area to develop chatbots that address clients' most common questions, direct them to solutions for specific assets, guide them through administrative tasks, assist with debugging, gather the necessary data to open a support case, and more.
The chatbots understand natural language to know exactly what product the client is asking about – or if they have a general question about how to interact with IBM support. They can then interact with the client to narrow in on the specific topic of interest so that the right answer is provided.
While some answers are curated, we also use Retrieval-Augmented Generation (RAG) to broaden the range of questions that can be addressed. The chatbots combine user-provided information with product-specific knowledge to conduct targeted searches within knowledge content, which is then used to prompt a large language model (LLM) to generate a response for the client, along with relevant references.
However, answer quality depends heavily on the chatbot’s ability to accurately interpret user intent, its defined topic scope, retrieval quality, and the capability of the generative AI to provide accurate answers without hallucinations. Our chatbots leverage watsonx technology (assistant, AI, data) along with SME-designed conversation flows to meet this goal.
The Virtual Assistant is available on IBM's Support Portal and I’d love for you to give it a try. Log in for the best experience.
Beyond How-to
Getting specific answers to questions with known solutions is great, but we can do more. If we connect the chatbot to our Cognitive Support Platform (CSP) and other TLS support infrastructure, we can start to get personalized value. Once we can detect the user intent is about a problem that requires us to go deeper, we can then guide them to more than simple answers.
For example: A system admin – let’s call him Sam - has an error code on his FlashSystem 9500 which he wants to investigate.
• He uses the IBM chatbot to ask about the error, its meaning, and possible corrective actions.
• The chatbot provides a summary of the issue and potential impacts and actions.
• Sam reviews the information but, in the end, determines that the issue is still not resolved.
• The chatbot then provides Sam guidance for the next steps for opening a case and collecting the right data for the case to be worked by a Support Agent.
• The chatbot will even be able to open the case for Sam, assist with log upload, and provide automated log analysis.
• Ultimately, the analysis may show that Sam’s system is missing a critical code patch which is why this error has surfaced. The automated analysis can then update the case with this information and instructions for how to resolve the error and update the code.
• Sam has now resolved an issue with the assistance of the IBM chatbot and other automation, on his own, and with less effort and time than he was previously able to do.
Now, not all problems will take this happy path. However, if the problem does require human interaction, the virtual assistant has collected all the information required, performed initial problem determination, and collected any remedial steps already taken.
The virtual assistant can hand the case off to a human support agent with this work fully documented in the support case, therefore Sam does not need to open a case and start over. In fact, the virtual assistant will notify the IBM support agent when the case is opened so they can contact Sam when they have performed their analysis.
The journey continues…
This is just the beginning. What we have described so far is all reactive, but we can also use virtual assistants as the first point of entry for clients looking for proactive solutions and insights. They can offer suggestions for maintenance best practices, vulnerabilities, and alerts for support contract renewals. The possibilities are endless. Join our TechXchange community and stay tuned for the upcoming news!
#PremiumSupportServices#InfrastructureServices