Feature Articles

Tech Talk: Hang-on! I’m talking to a what - The rise of the bot

By Ken Wong - 2 Jul 2023

Tech Talk: Hang-on! I’m talking to a what - The rise of the bot

Note: This feature was first published on 20 June 2023.

Wiz.AI’s CEO and Co-Founder, Jennifer Zhang. Image source: Wiz.AI.

The core technologies that go into a chat or talkbot 

Chatbots are all the rage these days. In its broadest terms, a chatbot is a platform that uses Artificial Intelligence (AI) and natural language processing (NLP) to understand customer questions and automate responses to them, simulating human conversation. AI for Customer Service if you will. 

As we look at the excitement around chatbots and Artificial Intelligence, and to better understand what an AI chatbot does, how intelligent its responses can be, and with over 95% of people fooled by it into thinking they’re talking to a real person, we spoke to Wiz.AI’s CEO and Co-Founder, Jennifer Zhang to find out more.

Chatbot vs. Talkbot: What's the difference? 

Wiz.AI deploys what the company calls a Talkbot, as opposed to a chatbot. According to the company, the key defining feature that differentiates the Talkbot from the chatbot is the Talkbot's ability to build a stronger relationship between the customer and your business. The company’s Talkbots can listen, understand, and speak in different ASEAN languages with localised accents. It can even incorporate human-like actions such as pauses, clarification, and chasing statements. Talkbots are able to embed the client’s information into their responses and this would facilitate a more personalised and realistic experience.
 

What does an AI bot do? 

At WIZ.AI, our AI Talkbots are expertly designed to deliver an outstanding customer service experience for businesses across a diverse range of industries, including banking, finance, insurance, FMCG, Ecommerce, healthcare, and telecommunications. Powered by our enterprise Large Language Model (LLM) with Natural Language Processing (NLP) and Dialogue Management, and coupled with a state-of-the-art speech engine, WIZ.AI Talkbots produce highly localised engagements. In fact, over 95% of callers are unable to distinguish between a human agent and our AI-powered Talkbot.

Our Talkbots are capable of handling an array of tasks, such as offering payment reminders and payment options, scheduling appointments, and even processing orders. By automating these functions, our bots enable businesses to conserve time and resources while simultaneously enhancing customer satisfaction and engagement.

Multi-point integration with telephony, WhatsApp, SMS, and email provides a seamless omnichannel experience for customers, ensuring they receive consistent and efficient service across all communication channels.

Where can an AI bot be installed and integrated with our CRM system or other business software/platform?  

Image source: Pexabay.

Our enterprise Large Language Models (LLMs) can be deployed on our cloud, the clients' private cloud, or even on-premises for those clients who require additional security. 

WIZ.AI's AI Talkbots and the like, excel in engaging customers through telephony and text-based channels such as WhatsApp, SMS, and email, delivering a truly omnichannel experience. This multi-point integration ensures seamless, targeted customer interactions at the right time with the right message. 

Talkbots can seamlessly integrate with a wide range of CRM systems and other business software or platforms. In addition to popular CRM systems like Hubspot, Salesforce, Zendesk, and Microsoft Dynamics, our solutions can also integrate with industry-specific CRM systems, such as debt management systems used by banks and financial institutions.

Once the integration is set up, the Talkbot can be programmed to perform various tasks and functions. These include retrieving customer data from the CRM system, updating customer records, scheduling appointments or follow-up activities, and more.

The AI bot can have various forms and provide statistics on customer behaviour. Does it collect users’ personal data? 

No. WIZ Talkbots will not collect user's personal data. During the conversations, some tasks may require the user to input some key information like birthdays, or the last 4 digits of some IDs. The WIZ system ensures that the data is encrypted and sent to our clients' internal system for authentication purposes. WIZ.AI will never store or keep this info – all customer data stays with our clients. We take data security and privacy very seriously.

When clients engage us to develop an enterprise LLM, we use their in-house proprietary data to train the AI engine. That information remains the property of our clients and we only tap into their resources for development and maintenance or updating the LLM and Talkbot. The enterprise LLM we develop is also the property of the client, and therefore they have full control over the data that resides in the LLM.

We are ISO 27001 certified, which is a globally recognised standard for information security management systems (ISMS). This certification demonstrates our commitment to the highest levels of data security and privacy. In addition to our ISO 27001 certification, we strictly adhere to data protection regulations and guidelines, such as GDPR, CCPA, and other regional data privacy laws.

How do I reconcile that with the PDPA?

Image source: Unsplash.

At WIZ.AI, we fully understand the importance of complying with data protection regulations, including the Personal Data Protection Act (PDPA). Our AI Talkbot solutions are designed with data privacy and security at their core, ensuring that we adhere to all applicable legal requirements.

We work closely with our clients to ensure that they deploy our Talkbots only for users who have provided the necessary consent. In addition to obtaining user consent, we also ensure that personal data is protected from unauthorised access, use, or disclosure. This includes implementing appropriate security measures to protect against data breaches and ensuring that personal data is not used or disclosed for purposes. Our ISO 27001 certification highlights our dedication to maintaining the highest levels of data security and privacy.

When should an AI chatbot be launched?

At WIZ.AI, we believe that an AI Talkbot should be launched when the organisation is ready to enhance customer engagement, optimise resource allocation, and improve overall efficiency. However, there are several factors to consider before launching an AI Talkbot:

  • Clear objectives: Identify the specific goals and objectives our client wants their Talkbot to achieve. This can include providing customer support, answering frequently asked questions, booking appointments, or generating leads.
  • Data readiness: Ensure that our client has sufficient data and information for the Talkbot to operate effectively. This includes having a well-structured knowledge base, understanding their customers' needs, and being able to anticipate customer questions.
  • Integration capabilities: Ensure that our client's existing systems, such as CRM platforms or telephony solutions, can be seamlessly integrated with the AI Talkbot for a smooth and efficient customer experience.
  • Compliance with regulations: Ensure that the Talkbot adheres to all relevant data protection regulations, such as GDPR, CCPA, or PDPA, depending on our client's region of operation.
  • Continuous improvement: Be prepared to invest time and resources into training and refining the AI Talkbot's capabilities, as well as monitoring its performance to ensure continuous improvement.
  • Internal readiness: Make sure that our client's team is prepared to support the Talkbot's deployment, including addressing any technical or operational challenges that may arise.

Can an AI bot replace my website? 

Image source: Pexabay.

We view AI bots as valuable additions that complement and enhance our client's customer engagement strategy, rather than as replacements. Websites are essential for establishing an online presence, offering comprehensive information about their products, services, and company background, and driving online traffic and lead generation. Talkbots, conversely, are designed to deliver interactive, personalised customer experiences through real-time conversational interfaces, assisting with tasks, answering questions, providing recommendations, and even handling transactions.

When integrated into their customer engagement framework, AI bots can offer numerous benefits. These can be improved customer engagement through instant support and assistance, reduced response times by handling multiple queries simultaneously, or enhanced efficiency by automating repetitive tasks and freeing up their team to focus on more complex or high-value tasks. 

So while AI bots can significantly improve our client's user experience, they should not be considered a replacement for a well-designed and informative website. Instead, they should be regarded as a vital component that enhances customer engagement and satisfaction.

Our business does not have a website. Can we use an AI bot? 

WIZ.AI's AI Talkbot can be an excellent solution for businesses without a website. In fact, some of our clients are Small and Medium Businesses (SMBs) that do not have a website but use our Talkbot to automate their customer engagement, such as appointment reminders, promotions, and even lead generation. We developed our solution, TalkGPT, specifically to cater to their needs.

TalkGPT harnesses the power of generative AI and our proprietary large language model (LLM) to enable businesses to rapidly launch customer engagement campaigns in just 5 minutes with a few clicks. Clients simply input their company profile, objectives, and target market, and TalkGPT will generate a highly customised omnichannel campaign including scripts and messages. The client can then easily review, choose which option is most appropriate, including the local language version, and, if needed, customise it themselves.

TalkGPT is designed to provide authentic and effective communication, natural human-like conversations, and high scalability and efficiency for transformative customer outreach. With TalkGPT, businesses without a website can still create effective customer interactions and dialogue flows in less time, making it an ideal solution for SMEs looking to improve their customer engagement.

Are AI chatbots, AI digital assistants, and AI virtual assistants the same?

Image source: Wiz.AI.

AI chatbots, AI digital assistants, and AI virtual assistants are similar in that they all use artificial intelligence to interact with users and provide support or assistance. But there are differences in the way they are used and deployed.

An AI chatbot is typically a text-based interface that uses natural language processing (NLP) and machine learning algorithms to understand and respond to user enquiries. Chatbots are often used for customer service and support, helping users find information or complete simple tasks.

An AI digital assistant is a more advanced type of chatbot that can handle more complex tasks and interactions. They also often rely on voice-based commands, like what we have with Siri, Alexa, or Google Assistant. Digital assistants are designed to perform a wide range of tasks, such as scheduling appointments, making reservations, and sending messages.

An AI virtual assistant is similar to a digital assistant but is typically more personalised and focused on the individual user. Virtual assistants may provide personalised recommendations or assistance, such as suggesting nearby restaurants or providing reminders for upcoming events, based on the user's preferences, habits, and activity history.

Can I have a bot in more than one language? Can it handle accents?

WIZ.AI Talkbots are actually designed to support multiple languages and dialects. We have a Singlish Talkbot, a Bahasa Talkbot, and a Taglish Talkbot, to name a few. 

When we started, we decided to focus on Southeast Asia because we know that there aren't any big tech companies that serve this region of over 1,000 languages and accents. So our bots are unique in that they can understand accents and even local expressions that an English-language bot designed using a general large language model cannot. For example, in Thailand, our clients often ask whether WIZ Talkbots can distinguish and speak using a Northern Thai and a Southern Thai accent. There are subtle differences which their customers can pick up, and our Talkbot technology addresses those small details.

What determines how smart an AI bot is? 

Image source: Pexels.

The "smartness" or intelligence of an AI bot is determined by its ability to understand and respond to user input in a way that is both accurate and helpful. This ability is driven by a combination of factors, including the quality and quantity of training data, the algorithms and models used to process that data, and the quality of the user interface and experience.

The quality and quantity of training data are critical factors in determining the intelligence of an AI bot. To perform well, an AI bot must be trained on large datasets of conversational data that are relevant to the specific task it is designed to perform. The more diverse and representative the training data is, the more accurate and robust the bot's responses are likely to be.

The algorithms and models used to process the training data are also important. AI bots typically use machine learning algorithms and models that are designed to process and analyse large amounts of data, recognise patterns, and make predictions. The quality and sophistication of these algorithms and models can significantly impact the bot's ability to understand and respond to user input.

Finally, the quality of the user interface and experience is also important in determining the intelligence of an AI bot. A well-designed user interface can make it easier for users to interact with the bot, provide more accurate and helpful responses, and generally enhance the user experience.

What determines AI bot functionalities?

The functionalities of an AI Talkbot are determined by several factors that contribute to its overall capabilities and performance. These factors include:

  • Spoken Language Understanding: The Talkbot's ability to comprehend and process spoken language effectively plays a significant role in determining its functionalities. This enables the Talkbot to engage with users in a more natural and conversational manner.
     
  • Domain-specific knowledge: AI Talkbots built on WIZ.AI's enterprise LLM have a wealth of domain-specific knowledge, which enables them to provide accurate and relevant information to users. This tailored knowledge enhances the Talkbot's capabilities in specialised areas and industries.
     
  • Human-AI collaboration: The AI Talkbot is designed to handle routine tasks efficiently, while also being smart enough to recognise when a human touch is needed. In such cases, the Talkbot routes high-value calls to human agents, enabling these agents to focus on more complex issues and provide personalised assistance. This collaborative approach enables both the Talkbot and human agents to
    work together effectively, optimising customer service and support.
     
  • Omnichannel customer engagement: The ability of AI Talkbots to engage with customers across various channels, such as telephony, WhatsApp, SMS, and email, significantly impacts their functionalities. This seamless integration allows for a more cohesive and consistent customer experience.
     
  • Instant scalability: A key determinant of AI Talkbot functionalities is their ability to scale rapidly to accommodate changing business needs or increased customer interactions. This ensures that the Talkbot can efficiently handle large volumes of queries without compromising on quality or responsiveness.

What happens if the AI bot does not give an answer to a question or gives an offensive response? 

Image source: Wiz.AI.

If an AI Talkbot encounters a question it doesn't know the answer to or is unable to provide an accurate response, there are a few strategies it can employ to handle the situation:

  • Escalate to a human agent: The Talkbot can be programmed to recognize its limitations and transfer the conversation to a human agent who can provide the necessary assistance. This ensures that customers receive accurate information and support when needed.
     
  • Provide a general response: If a specific answer is unavailable, the Talkbot can offer a general response, guiding the user to relevant resources or suggesting alternative ways to find the information they seek.
     
  • Admit limitations: In some cases, it might be best for the Talkbot to admit that it doesn't have the answer and apologise for the inconvenience. This honest approach can help manage user expectations and maintain their trust. 
     
  • AI Talkbots continuously learn and improve through user interactions.

    When a Talkbot fails to provide an accurate answer, this feedback can be used to enhance its knowledge and performance in future interactions. By employing these learning strategies, AI Talkbots can maintain a positive user experience, even when faced with questions they cannot answer. Unlike human agents, Talkbots never have a bad day and never take things the customer says personally. This ensures consistent, professional, and respectful interactions with customers at all times. In the unlikely event that an offensive response does occur, we encourage clients to provide feedback so that we can quickly address the issue and update the Talkbot's knowledge and performance, to continually improve the AI Talkbot experience for all users.

Can there be more than one answer to a question in the conversational AI module?

By understanding the nuances of user inquiries, our Talkbots can generate different answers tailored to address specific aspects of the question or offer various suggestions based on available information.

The flexibility of our Talkbots in providing multiple answers allows them to deliver a more engaging and interactive experience for users, ensuring that their questions are comprehensively and accurately addressed. Moreover, our AI systems continuously learn and evolve from user interactions, which contributes to the ongoing improvement of the quality and relevance of their responses.

We can also upload files and links to the AI Talkbot, enabling it to deliver a more comprehensive and interactive user experience. Our AI Talkbots can be configured to handle various types of media and content, including images, documents, videos, and links to relevant websites or resources. This enables the AI Talkbot to provide users with a wealth of information, answering their questions more effectively and offering them additional resources to explore.

For example, after the call, the Talkbot can follow up with a WhatsApp message containing a link to the sign-up page or promotional offer. 

What are some of the challenges that I can expect with a Talkbot?

Image source: Pexabay.

Integrating an AI Talkbot into our client's business can bring numerous benefits, but it's important to be aware of potential challenges as well. Here are some common challenges:

  1. Pinpointing the perfect touchpoints: Sometimes clients come in with a pre-set notion of where they want to deploy a Talkbot. However, identifying the most effective touchpoints for a Talkbot requires analysing a customer's engagement strategy in detail. This ensures a seamless customer journey and optimal results.
     
  1. Tailoring and training: Customising the Talkbot to meet a company's specific needs and training it on industry-specific knowledge may require several iterations. This process guarantees that the Talkbot comprehends your business's intricacies and performs at the desired level.
     
  1. Encouraging user adoption: Motivating the staff internally to fully embrace the Talkbot's capabilities and benefits can be a challenge. Promoting its features and fostering user engagement with the Talkbot is key to unlocking its full potential.

How do I measure the success or failure of the chatbot?

Measuring the success of Talkbots can be directly tied to their impact on business results, focusing on the tangible outcomes and return on investment (ROI) they bring to your organisation. Here are some key performance indicators (KPIs) that can help you evaluate the effectiveness of your Talkbot in achieving business objectives:

  1. Appointment rate: Calculate the number of appointments secured by the Talkbot compared to the total number of user interactions, to understand its effectiveness in booking appointments for your business.
     
  1. Debt collection rate: Measure the percentage of debt successfully collected by the Talkbot, indicating its efficiency in assisting with debt recovery efforts.
     
  1. Return on Investment (ROI): Analyse the overall financial impact of implementing the Talkbot by comparing the revenue generated or costs saved against the investment made in deploying and maintaining the system.
     
  1. Lead generation: Assess the number of new leads acquired through interactions with the Talkbot, highlighting its ability to attract potential customers and drive business growth.
     
  2. Conversion rate: Track the percentage of users who complete a desired action after engaging with the
    Talkbot, such as making a purchase, signing up for a newsletter, or enrolling in a program, to determine its impact on your business goals.

By focusing on these results-driven KPIs, you can more accurately evaluate the success of your Talkbot and its contribution to your organisation's bottom line. Continuously monitoring these metrics and making data-driven optimisations will help ensure your Talkbot remains an effective and valuable tool for achieving your business objectives.

Join HWZ's Telegram channel here and catch all the latest tech news!
Our articles may contain affiliate links. If you buy through these links, we may earn a small commission.