chatbots switzerland; chatbots-lausanne; human-chatbot; build-chat-botsManuel has been developing apps for 18 years in areas like mobile apps, bots and IoT. He’s currently the CTO of Coteries.

We started developing Chatbots for Slack and Telegram and then we went for Facebook, which was announced 3 months later.

We went deeper on Facebook after they announced their SDK (software development kit) to make bots on Messenger. First, we made some prototypes just to test how it was going to be and we realised, fast enough, that the UX was not the same that we are used to with websites and all the mobile apps that we do.

It was very different because it is a driven conversation, and sometimes people don’t get that. Everyone is so used to speaking to humans, that sometimes they think the bot is one… but in reality, it’s not like that. Robots have to be taught about what to say, what answers to give, so they can have a fluent conversation. However, there is a conversational line to respect, and when people go out of that line, bots don’t work as well as humans (yet!).

Recent news: “Facebook robots developed their own language”

Ok, let’s not freak out about this! You can’t believe all the headlines you see out there. What happened has a logical explanation, to put it simply, these bots haven’t created a new language, at the most, they gave some grammar mistakes! FastCompany did an extraordinary job explaining all of this. 

This Facebook fever originated big headlines in the news as if doomsday was coming, but relax… not yet. We are still far from that reality by now.

To add, we have to do a job of educating people before starting the conversation. This means giving them buttons with pre-made questions or some hints about what fields can the chatbot cover.

2 Reasons why it’s extremely important to direct the conversation between humans and bots:

  1. AI (Artificial Intelligence) is not strong enough to know every answer;
  2. Humans want to have human conversations, and if you don’t guide the talk, it becomes a mess.

When this happens normally we tell the bot to send a message saying it couldn’t understand the question.

chatbots ; lausanne-chatbot; switzerland-developers; coteries-chatbots

Will we be able to have a normal conversation with bots in the future as we have with humans today? The reality is, we don’t know. By now, there are a lot of things that machines cannot understand. Also, for now, they need humans to help them learn.

Developing chatbots is not easy:

Basically, you send a sentence and the AI takes this text and cuts it in words/verbs. This how it knows what actions must be taken or what is the context.

Languages normally bring the biggest problems while developing a chatbot. One time we had a problem with the city of Geneva because in Switzerland this city can be written in 4 different languages. Swiss German is also a challenging issue to overcome while developing.

We won’t say it’s impossible to create your own chatbot online or by yourself, but it’s quite challenging if you aim to have a decent outcome. Good products demand dedication and time to be ready. For simpler bots you can find several platforms that do the hard work for you, but if you need something more complex, that’s a different story.


Liked this article? Share with your friends on social media: