An adaptable AI chatbot that gets it right the first time
IBM Watson Assistant is built on deep learning, machine learning, and natural language processing (NLP) models to understand questions, find or search for the best answers, and complete the user’s intended action through conversational AI. A key component is the large language model (LLM) that has been trained on an enormous dataset—including millions of business-related words and phrases—to increase understanding based on context, enable accurate information extraction, deliver granular insights from business documents, and boost the accuracy of responses. Watson also uses intent classification and entity recognition to better understand customers in context and transfer them to a human agent when needed.
Proven up to 14.7% more accurate than competitive solutions in a recent published study on machine learning.
Achieves the same or better accuracy 50 times faster (link resides outside IBM) (PDF, 240 KB) on average.
The intent detection algorithm is now 79% accurate at answering customer requests on its own in real time.
Build an AI chatbot that works better
Make it easy for customers to complete more actions in the fewest steps possible, while speaking in their own words with their own quirks.
AI chatbot that understands
Understands any request
Best-in-class NLP and LLM can be quickly trained to understand a new topic in any language with only a handful of example sentences.
Adapts to your domain
Deep learning models automatically adapt to your business' domain based on the sentences you provide as training data.
Knows when not to answer
Irrelevance detection models help the system know when to “buzz-in” confidently or when to pass to help documents or a human agent.
Recognizes plain-language responses
Powerful entity detection models can recognize plain-language responses from your customers like synonyms, dates, times, numbers and more.
Doesn't ask redundant questions
Reduce frustration by using information gathered in previous requests to skip steps and streamline the conversation.
Handles natural conversation without breaking
Gracefully handle vague requests, topic changes, misspellings, and misunderstandings during a customer interaction without any additional setup.
Find answers in existing content
Powered by IBM Watson Discovery, this intelligent search capability finds accurate and concise answers to customer inquiries in any existing documents, websites, and knowledge bases—even if not explicitly trained with a specific intent or action. A foundation of powerful large language models (LLMs) enhances understanding.
Optimize and scale accurately
Watson Assistant automatically clarifies vague requests and uses your customers' selections to improve its understanding going forward.
AI that finds insights
Watson Assistant uses machine learning to identify clusters of unrecognized topics in existing logs helps you prioritize which to add to the system as new topics.
Sometimes existing topics need more diverse training examples. That's why Watson Assistant recommends sentences that you should add to existing topics.
Intent conflict resolution
Automatically detects and alerts you of potential overlaps in your training data which would negatively affect the performance of your assistant.
Frequently asked questions
Does chatbot technology use machine learning or natural language processing (NLP)?
Watson is built on deep learning, machine learning and natural language processing (NLP) models to elevate customer experiences and help customers change an appointment, track a shipment, or check a balance. Watson uses machine learning algorithms and asks follow-up questions to better understand customers and pass them off to a human agent when needed. In addition, Watson leverages large language models (LLMs). These foundation models from Watson Natural Language Processing (NLP) deliver advanced processing and understanding of text, enabling the accurate extraction of information and insights from business documents, accelerating processes, and generating insights. The large language models (LLMs) from IBM are explicitly trained on large amounts of text data for NLP tasks and contain a significant number of parameters, usually exceeding 100 million. They facilitate the processing and generation of natural language text for diverse tasks. Each LLM model has its strengths and weaknesses and the choice of which one to use depends on the specific NLP task and the characteristics of the data being analyzed.
Try out the enhanced intent detection model. This new model, which is being offered as a beta feature in English-language dialog and actions skills, is faster and more accurate. It combines traditional machine learning, transfer learning and deep learning techniques in a cohesive model that is highly responsive at run time. For more information, see Improved intent recognition.
What is conversational AI?
Conversational artificial intelligence (AI) refers to technologies, like chatbots or virtual agents, which users can talk to. They use large volumes of data, machine learning, and natural language processing to help imitate human-like interactions, recognizing speech and text inputs and translating their meanings across various languages.
What is generative AI?
Generative AI refers to deep-learning models that can generate text, images, audio, code, and other content based on the data they were trained on. Watson Assistant’s conversational AI platform can use generative AI capabilities to better understand customer needs, automate self-service and answers, and deliver exceptional digital experiences.
What are common chatbot platform use cases?
Customer care is the most common chatbot use case. Chatbots are helpful to both product- and service-based companies looking to provide a superior user experience by to answering customer questions, guiding customers through simple troubleshooting, and connecting customers to the resources they need.
Chatbots are also often used by sales teams looking for a tool to support lead generation. Chatbots can quickly validate potential leads based on the questions they ask, then pass them on to human sales representatives to close the deal.
Chatbots can even be used in e-commerce by acting as a digital sales clerk, akin to what customers would experience in brick-and-mortar stores. E-commerce chatbots can provide a personalized shopping experience that converts passive visitors into engaged prospects.
How do you use chatbots to automate customer support workflows?
A customer browsing a website for a product or service may have questions about different features, attributes or plans. A chatbot can provide these answers, helping the customer decide which product or service to buy or take the next logical step toward that final purchase. And for more complex purchases with a multistep sales funnel, the chatbot can qualify the lead before connecting the customer with a trained sales agent.
How do chatbot solutions improve customer satisfaction?
Today, chatbots can consistently manage customer interactions 24x7 while continuously improving the quality of the responses and keeping costs down. Chatbots automate workflows and free up employees from repetitive tasks. A chatbot can also eliminate long wait times for phone-based customer support, or even longer wait times for email, chat and web-based support, because they are available immediately to any number of users at once. That’s a great user experience—and satisfied customers are more likely to exhibit brand loyalty.
Do virtual agents or chatbots respond to customers in real-time?
A chatbot can answer questions 24 hours a day, seven days a week. It can provide a new first line of support, supplement support during peak periods, or offer an additional support option. At the very least, using a chatbot can help reduce the number of users who need to speak with a human, which can help businesses avoid scaling up staff due to increased demand or implementing a 24-hour support staff.
What is an API?
An API is a software intermediary that enables two applications to communicate with each other by opening up their data and functionality. App developers use an API’s interface to communicate with other products and services to return information requested by the end user. When you use an application (such as a virtual assistant) on your phone or computer, the application connects to the Internet and sends data to a server via an API. The API then helps the server interpret the data so it can perform the necessary actions. Finally, the server sends the requested data back to your device via the API where it is interpreted by the application and presented to you in a readable format. Without APIs, many of the online applications that we’ve come to rely on would not be possible.
Can you interact with a customer service chatbot on a mobile app?
Yes, in fact deploying chatbots to mobile apps is a common use case. Lloyds Banking Group, the UK’s largest retail bank, has multiple virtual assistants, most prominently a retail mobile banking app where more than 10 million mobile customers can communicate with the bank at a time that suits them.
Can I deploy my AI bot to social media channels like Facebook Messenger, Whatsapp, Slack, or Amazon Alexa?
Yes, you can deploy your chatbot to Facebook Messenger, Intercom, Slack, SMS with Twilio, and WhatsApp. You can even deploy to Amazon Alexa. Learn more about Watson Assistant integrations.
Are there chatbot templates to use?
By default, the web chat window shows a home screen that can welcome users and tell them how to interact with the assistant. For information about CSS helper classes that you can use to change the home screen style, see the prebuilt templates documentation.
Can I connect my AI chatbot software to a knowledge base for FAQs?
Watson Assistant's Search Skill provides accurate answers to customer inquiries in any existing documents, websites, knowledge bases and enterprise applications, including Salesforce, SharePoint, Box and IBM Cloud Object storage.
How does the Watson Assistant dialog drag-and-drop editor work?
With most tools, fixing a conversation flow is brittle and error prone, but with Watson Assistant's drag-and-drop editor, you can quickly change your content, conditions, or step prioritization without worrying about causing more problems.