From messages to conversation
Posted: Mon Dec 23, 2024 7:20 am
d. Help, I can't get the WiFi to work on my flight.
2. Where will the bot be placed? Only Facebook Messenger or own site, app? 'Where our customers are' as a basis. Integration of bots in its own site and all social channels, such as Facebook and WeChat (China).
3. What information sources will it use: APIs, business systems, add-ons, etc? Integration into back-end booking and transactional systems, for direct booking, changes and payments.
4. When does it escalate to a human? In general: after a question has been asked 3 times, but the bot doesn't know the answer.
5. How are you going to build it? What engine are you going to use? Various, including API.ai (see 'Building your bot' below for options).
6. What voice/personality will it have? Neutral, in line with the experience a customer would have with a live agent. It's about a consistent customer experience.
1 to 6 of 6 resultsPreviousNext
Building your chatbot
In the development of a more complex chatbot you will hospital mailing list see that your bot engine is linked to your business back end systems (for example to SAP to retrieve certain customer data) and to a chat platform.
Popular bot engines and authoring tools to get started with your bot include API.ai (Facebook Messenger, Slack and Google Conversations), WIT.ai (Facebook), Chatfuel, Pullstring, Botkit Studio, Motion.ai and Botsify.
In API.ai, a bot can be created in a few simple steps, integrated with Facebook Messenger. You can create various intents . These represent a certain situation with a number of potential user questions, which should lead to an action. The nice thing is that if the user question is not exactly the same as programmed, API.ai still recognizes it. The next step is to create conversation trees, with which much more complex bots can be created.
It seems easy to launch a bot in an hour and a half. However, there is still a long way to go before a bot comes close to a decent conversation, without too many errors. A stronger AI can help to at least reduce the manual scripting work. Try it yourself with API.ai for example and let us know how you get on in the comments below.
2. Where will the bot be placed? Only Facebook Messenger or own site, app? 'Where our customers are' as a basis. Integration of bots in its own site and all social channels, such as Facebook and WeChat (China).
3. What information sources will it use: APIs, business systems, add-ons, etc? Integration into back-end booking and transactional systems, for direct booking, changes and payments.
4. When does it escalate to a human? In general: after a question has been asked 3 times, but the bot doesn't know the answer.
5. How are you going to build it? What engine are you going to use? Various, including API.ai (see 'Building your bot' below for options).
6. What voice/personality will it have? Neutral, in line with the experience a customer would have with a live agent. It's about a consistent customer experience.
1 to 6 of 6 resultsPreviousNext
Building your chatbot
In the development of a more complex chatbot you will hospital mailing list see that your bot engine is linked to your business back end systems (for example to SAP to retrieve certain customer data) and to a chat platform.
Popular bot engines and authoring tools to get started with your bot include API.ai (Facebook Messenger, Slack and Google Conversations), WIT.ai (Facebook), Chatfuel, Pullstring, Botkit Studio, Motion.ai and Botsify.
In API.ai, a bot can be created in a few simple steps, integrated with Facebook Messenger. You can create various intents . These represent a certain situation with a number of potential user questions, which should lead to an action. The nice thing is that if the user question is not exactly the same as programmed, API.ai still recognizes it. The next step is to create conversation trees, with which much more complex bots can be created.
It seems easy to launch a bot in an hour and a half. However, there is still a long way to go before a bot comes close to a decent conversation, without too many errors. A stronger AI can help to at least reduce the manual scripting work. Try it yourself with API.ai for example and let us know how you get on in the comments below.