Making simple chatbots better

Written by Livio Pugliese

February 6, 2024

Many deployed chatbots are far from holding real conversations. But they too can be enabled for fluent dialog. This is how we do it.

When you consider chatbots these days you think of ChatGPT, Google Bard, Bing Chat, etc. These are all based on Large Language Models (LLM) and are able to answer pretty much any question that users may ask. But in fact, there are thousands of deployed chatbot that help users with customer service issues every day, which are way more limited. Many chatbots in use today have simple interfaces with predefined questions that users can select by typing keywords or clicking on buttons. They work, sometimes they work well, depending on the domain and the type of business they serve. But there is no question that using these chatbots for voice interactions would result in a bad customer experience. That is, unless we at Interactive Media intervene.

In this article I will describe how to enable even these chatbots for voice, with excellent customer experience.

Type of chatbots

According to a blog post by IBM, there are four categories of chatbots:

1. Menu or button-based chatbots. These are the simplest chatbots, very much deployed on web pages, that guide the user through explicit choices. They are equivalent to the traditional tone based IVR for voice and can work well if the domain is simple and the choices are clear. Obviously, if the user needs something that is not included in the menu, they can’t help and should refer the customer to a human representative.

2. Rules- and keywords-based chatbots. These chatbots let the customer ask questions in somewhat free format, then match keywords in the question with their knowledge base and present text that contains those words. They are in essence interactive FAQ reading tools. The problem here is that if the question is complex, these chatbot can’t answer it and should forward the interaction to a human.

3. AI-powered chatbots. These chatbots have Natural Language Understanding (NLU) and Natural Language Processing (NLP) capabilities and can handle dialogs with multiple exchanges with users. They are based on AI engines acting on a knowledge base that is tailored for the specific domain they are serving. This means that, users can ask any question about information in the domain, and the chatbot will understand the question and answer according to its knowledge. Sometimes the chatbot will ask a clarification question if the user’s initial question is ambiguous. However, if the user asks something outside the chatbot specific knowledge base, the chatbot will not be able to answer.

4. LLM-based, generative AI chatbots. Well, these chatbots have been all the rage in the past year or so. They are fluent and can answer any question, since their knowledge base is enormous, they can also create new answers by putting together information and text in a statistical way. They are the future, and the high-end present, of customer service, as one of their applications. But there are many others…

If you read about chatbots in magazines and specialized websites, you’ll find 90% of the articles about the 4th type, maybe still 10% about the 3rd type, and nothing about #1 and #2. But many #1 and #2 chatbots are still serving customers – and will for years to come: they are paid for, and 3rd and 4th type chatbot applications for customer service are expensive.

Where simple chatbots can do the job well

Sometimes you don’t need sophisticated AI chatbots to serve your customers well. If the customer service domain is simple, the questions asked are for the most part always the same, and you want users to feel at ease with clicking buttons or following menus, a simple, Type 1 chatbot will do just fine.

For example, banks have a limited number of services that they can perform using chatbots. For this reason, often the UI consists of a series of buttons and menus where the user selects one item and continues to the desired service.

Another example may be purchasing a train ticket. You want the user to specify the date of the trip, the origin station, the destination, and the class of travel. Easy peasy. A chatbot with web widgets and keywords matching for cities will do the job.

How to enable dialogs and voice with simple chatbots (without excessive costs)

Even a simple chatbot like a type 1 has access to a valuable knowledge base that would be great to access by voice. And indeed, Interactive Media offers a service to add the voice channels to chatbots easily and conveniently, called PhoneMyBot. But a type 1 chatbot needs precise inputs from users which are not easily translatable into a voice conversation. Here we can use an add-on to PhoneMyBot, implemented using Interactive Media’s conversational AI platform, MIND. In essence, we add a very simple type 3 application in front of the main type 1 chatbot, able to figure out the intent of the call through a natural language dialog.

Once the intent of the call is determined, if it’s covered by the type 1 chatbot, MIND crafts the question in the format that the type 1 chatbot expects and sends it forward. PhoneMyBot will then retrieve the answer from the chatbot and speak it to the user. If the intent of the call is not in the chatbot’s knowledge base, MIND can determine to send the call to a human agent instead. This works with one chatbot, or several. Sometimes companies deploy more than one specialized chatbot for different tasks and MIND acts as aggregator and selector for the most useful chatbot.

This figure shows the architecture: PhoneMyBot and MIND exchange information initially, before connecting the caller to the chatbot.

For Interactive Media, this is a simple call steering application. We have developed many of these over the years and we can implement one fast and with contained costs. It’s a very cons-effective solution to add the voice channel to even a simple chatbot.

We would love to put more meat on the bone and talk to you about this solution: please contact Interactive Media at  or click the button below.

Other Articles

Other Articles

LLM-based chatbots and how to make them more reliable

LLM-based chatbots and how to make them more reliable

ChatGPT and its siblings are all the rage in customer service chatbots. This is fascinating and terrifying. How do we take the terror out of the equation?In the past year or so we witnessed an explosion of chatbots based on Large Language Models (LLM). The adoption of...

read more
Multimodal interactions: are they breaking through?

Multimodal interactions: are they breaking through?

Last week I watched a webinar and demo by a company providing tools and solutions for conversational customer service. Interactive Media, where I work, is in the same sector and I wanted to scoop out a competitor, see what they have and how they are presenting their...

read more
PhoneMyBot and ChatGPT: giving voice to AI

PhoneMyBot and ChatGPT: giving voice to AI

Talking with ChatGPT over the phone is cool. But we can also make it useful. In the past few months tech people worldwide have been talking almost exclusively about Open AI’s ChatGPT. It’s the first large language model chatbot to make a splash, and what a splash! It...

read more

Interact with us


Receive our exclusive content: