The better VoiceBot
IVR was yesterday. Today it’s AI.
With the help of artificial intelligence, our VoiceBot ensures that simple standard queries do not result in valuable agent time being blocked for the really important things in the customer dialogue:
The individual and high-quality help with advice-intensive concerns.
Your VoiceBot from telegra takes care of the rest. Quite simply.
The VoiceBot recognises the concerns of your callers on a semantic level with the help of artificial intelligence. It can give direct answers, initiate forwarding to other agent groups or perform database queries.
In many service centres, the share of recurring standard enquiries is between 20 and 30 percent. In future, our VoiceBot will be able to process these questions automatically and on a closed-case basis.
Automate the answering of consignment information on the phone in no time at all: Enter the consignment number and find out the consignment status directly and automatically. Very simple.
If you’re thinking about how much a voicebot will cost, you’re thinking about the wrong thing. Used correctly, this solution offers unique savings opportunities by automating your service processes. Admittedly: The above values are maximum values that customers may achieve. It’s best if we take a look at your very specific use case together.
The VoiceBot says exactly what you enter as text.
The VoiceBot recognises completely freely spoken concerns of the caller.
Multi-level IVR menus are replaced by a simple question and the direct recognition of the call request.
Simple routing: After recognising the call request, the system can forward to any destination (e.g. your agents) or give a direct answer.
Just get started: The system integrates into any existing ICT infrastructure without any effort.
All configurations and the entire administration are browser-based in HTML5.
In accordance with GDPR and from the German cloud from one of the leading service telephony experts on the market.
The system learns permanently and independently increases the rate of successfully recognised questions.
In future, your callers can simply formulate their call request freely. The VoiceBot ensures that the call is routed to the right place in your service centre. The caller’s concern is recognised with a high degree of accuracy. And if it doesn’t work out that way, you can define what should happen: Forward the call directly to an agent.
Many service centres are facing a high percentage of simple standard enquiries. In future, these enquiries can be automated and answered much more quickly at any time. This means that valuable resources can be used for conversations that are still best conducted person-to-person in the future.
The system transcribes spoken language into text in real time. This means that the contents of dialogues are now machine-readable. For the first time, you can also evaluate the content of conversation data, include it in statistics or display the content of upstream dialogue chains to the agent when answering a call.
The concept of the telegra VoiceBot is based on the fact that you can teach the machine to recognise certain concerns. To be clear from the outset: The system will not recognise requests that you do not define. So as a first step, think about what your callers want to know in most cases. Start with the standard questions. You can configure the system so that attachments that are not recognised are forwarded directly to an agent. After a certain period of time, you simply check your statistics to see which requests have not yet been recognised by the bot and define appropriate answers for these requests as well.
It is also possible to use the bot to query customer data and compare the results with a sub-system via http request. In the other direction, your sub-system can – depending on the result – “tell” the bot what it should do next.
In order to function as a concern recogniser and answerer, the system must first convert spoken language into machine-readable script. And it does this in real time. This text can then be further processed. The VoiceBot from telegra works amazingly well in transcription: even under adverse circumstances, when callers call from places with a lot of ambient noise (train stations, airports, means of transport, busy streets), the speech is converted into text with high accuracy.
Our VoiceBot says exactly what you enter as the answer text in the system. You simply think about what answer the bot should give when it recognises a request. Sounds simple, and that’s exactly what it is. How does it work? Our system offers a voice engine that provides a voice output of written text.
Within the configuration, you can also choose which voice should be used to address your callers and whether it should be a woman or a man. Do the voices sound familiar? Your callers will too: you can choose between Google and Amazon voices. They are familiar from Alexa and Google Assistant and ensure acceptance and recognition for the user of your VoiceBot right from the start of the call.
In the training centre you make your bot fit for understanding the concerns of your callers. In simple terms, our VoiceBot operates by calculating probabilities when recognising a request. If the calculated probability falls below a certain value, this means from the bot’s point of view that it is not quite sure and will ask again. However, it is also possible that the probability is so low that the bot does not ask any more questions, but forwards the request directly to an agent (for example, if a request has been made that has not yet been defined).
The training centre works in such a way that you can correctly assign requests that are either wrong or that were only recognised through enquiries. You help the machine a little and give it support. This causes an even faster learning curve in recognising the concerns of your callers. All conversations of the system are included and the filter function allows you to display exactly those cases in which the bot only worked poorly (i.e. by request) or not at all. Experience shows that 1-2 days of training the machine is enough to achieve recognition rates of more than 90 percent.
Statistics are the heart of every service centre organisation. The better you prepare the data, the easier it is for you to check the performance-relevant KPIs.
In the case of a bot, the crucial question is of course how good it was at recognising the callers’ concerns. Another important point is the question of what happened in the first place after the concern was recognised. How many calls could be answered directly, which calls were forwarded and where were they directed to.
You can evaluate all data on an annual, monthly or even hourly basis and go into the details of each call.
Our VoiceBot works in any ICT environment and with any telephone system or ACD in the world. But it works especially well with telegra Contact Centre software.
If you are interested in the following functions, you should take a look at the product page:
Find out about the advantages of using telegra’s Contact Centre software in combination with the VoiceBot.
No matter what stage you are at with your planning. We are your contact.