Chatbots — Myths & Reality of the upcoming revolution

“A bot is a robot, powered by AI, that can converse with you. It is designed to automate tasks that you would usually do yourself, like booking a table at a restaurant, a flight ticket or finding out a good place to go out at night. They can live inside an app, a website or messaging app.” — Recast.ai

Bot (or Chatbot) is the new buzzing word all over the tech press that few people really know about. For sure the tech guy from your company knows all about it. However, those who don’t work in a technical field often believe it is short for “robots”.

What I hope to do here is give you the basic knowledge you will need in order to understand this upcoming revolution.

This article is part of a series that I’m writing after attending Leade.rs Paris — a tech conference that was held by Loic Le Meur in April. Here is another article I wrote about the conference if you would like to know more about it.

As a reminder, and before diving into the subject, let’s talk about the difference between an app and a bot. As a user, you won’t need to download a bot ; you will interact with it directly through an existing platform such as Facebook Messenger, Slack or Telegram. Hence, a bot provider won’t need to build the whole infrastructure or even invest in UX/UI design. As a consequence, this solution will be less expensive than building an app.

The bot phenomenon is relatively new and is the logical follow-up of the virtual assistant era. The virtual assistant lacked intelligence while the bot can be smart. It is only the beginning and even if the press is crazy about the phenomenon, few large companies have invested in bots in 2016 ; many of them are in the Proof of Concepts phase.

** In case you don’t have much time and want to skip to the end,
I’ll explain briefly the different parts of the article here **

1. Most of the current available bots are lame
2. Five key rules to keep in mind as a bot creator
3. How to go one step further
4. What I saw at Leade.rs Paris

1. Most of the existing chatbots are lame because they are badly designed

“60% of the bots I see are focused on customer support.

But I’ve never had a good experience so far.” Phil Libbins

Even if the phenomenon is widely covered in the press, the reality is that bots are still at the beginning. Most of the bots available today are simply “dumb”. They work on IFTTT rules (If This Then That) and are rapidly out of words when you try and interact with them.

The challenge is now to leverage all the data that bots & apps have collected so far and to get smarter. I am certain that once people fully understand what bots can and cannot do, they will start using them, because it is designed to be simple, quick & useful. And because as humans, we are lazy.

The effort required to download a new app and test it out isn’t worth the download anymore. Users would rather have new services integrated into a messaging app they already use than to download yet another app.

“Over 2.5 billion people have at least one messaging app installed. Within a couple of years that will reach 3.6 billion, about half of humanity.” — The Economist

“Stop wasting money trying to pull people into your ecosystem. Push your content where your users are already active”. — Joe Toscano, Bots lover & Experience designer

What kind of bots can we see today in the market?

The most common fields were we can interact with bots are :
- Analytics
- Communication
- Customer support
- Education
- Entertainment
- Food & Places to go to
- Games
- Productivity
- And so on ...

If you want to go deeper, https://blog.recast.ai/2017-messenger-bot-landscape/

2. Five key rules to make your bot awesome

I am not going to give you technical rules such as coding, hosting & testing your bot, but rather five essential rules to keep in mind that I’ve learned at Leade.rs and through my research.

Be explicit about what the chatbot is & isn’t; about what it can, and cannot do

People must easily link your bot to something (a task, a theme, a personality) so that they come back. Otherwise, they will just try your bot once (assuming they have found it) and forget about it => Attraction vs Retention

Be explicit about the fact that it is automated — underline that it is not a human being

I realized this when making the demo of a bot I found to my mother. She didn’t get the fact that the bot was automated and it just puzzled her. Needless to say she did not like it. People need to understand they are talking to R2D2.

Create an identity that will interact well with your audience

A bot needs an identity as much as it needs a theme. It can be funny, neutral, sad or joyful as long as it is consistent with your target audience. The personality also comes from usage & data: learn how to interact with those who like your bot.

Focus on one use case and do it well

People like it when things are easy. Don’t try to complicate everything too soon. Find a target & a good use case and make it perfect.

Monitor the bot and improve it in the long run

Just like every other product, your bot should be improved over time and be adapted to its customers. Understand why people come, what they are looking for and refine it over time.

If you want to go deeper, https://blog.recast.ai/five-steps-bot-building/

3. How to go one step further — Contextual conversations

How to go beyond “one-off” use cases and get into the ‘contextual conversations’ to deliver the right level of personalization?

Bots are a new ideal alternative to communicate with a user database. Until now, brands could reach their users physically in stores or digitally by monitoring our clicks & other actions on the Internet. As for now, brands can listen to what we say / post / like / share everyday on social media. And tomorrow, thanks to bots, brands will directly interact with us, and more importantly, at scale!

The challenge is now to make us talk!

While we are used to talk to friends, family or colleagues, talking to a robot is still odd — even if it is well-written and has a personality we like. The challenge for the bot is to understand what we type, get our intents and respond accordingly. If the answer is adequate and contextualized, we might will be willing to respond and interact. If the answer is basic and makes us think of a standard call center, there is little chance that we will give it a try.

The ability to execute predictive models in real-time instead of relying on pre-calculated scores will be key in the future. No matter how accurate models that predict a customer’s behavior are, they risk being outdated if they are applied to outdated data. With bots, the model gathers the data directly, the conversation is easier and with the right level of intelligence, users will have the level of customer support they’ve been waiting for for a long time.

4. Bots we saw at Leaders

Replika

Replika goes one step further than most of the bots we can see out there. They are building an emotional bot, the imaginary friend you use to talk to when you were a kid.

Phil Libins said in Leade.rs - when talking to Replika’s founder - that if you want to order a pizza through a bot, that means you don’t want to talk and barely want to push a button. Basically, you just want a pizza. It’s a conversation that you are willing to spend money not to have thanks to technology. Most of the time, we act with this thought in mind.

Here, Replika goes three steps further! The idea is to interact on a regular basis, make you talk, free up your mind and chat with your Replika.

In the app store description, this is what you can see:

NB: Currently, you need an invite to talk to your Replika.

Jam

Jam is a French artificial intelligence bot that helps you in your daily life. You can ask Jam any question (find a place to go to at night, gather tips to get an internship, find a laundromat nearby, anything) and it will answer either with AI or with a human being if the AI is stuck.

Jam is one of the leading bot companies in France, with a great user database (essentially millenials) and with a successful business model. It was one of the few start-ups to be presented at the last Facebook F8 conference in 2017. Source

Citron

Citron is another French bot focused on finding nice places to go to in Paris. It’s main advantage is that Citron recommends places based on your current location.

It will celebrate its one year anniversary on June 29th at a Secret Party in Paris and it could be a good opportunity for you to meet them if interested— Facebook event here

Recast.ai

Recast.ai is not really a bot, but rather a collaborative platform to build, train, deploy and monitor intelligent bots for developers. They made a great impression on stage at the end of the two day conference to pitch the company when Jasmine Anteunis — Recast’s founder — was nominated in the Business Insider top 100 French entrepreneurs to watch.

By the way, they have a great blog on their website covering bots, AI and other tech subjects that I strongly recommend if you have some time to spare and are looking for information.

As we said earlier, building a bot is quite easy. You can even try and deploy your own personal bot in minutes using platforms such as Chatfuel. But the real challenge is actually to get users’ attention and to make them stay & interact with your bot.

As a final word, I would once again underline that language is what makes us human. It allows us to connect better, to convey emotions and feelings. This is what bots lack today and this is where creators are putting their efforts today. Stay tuned for more to come, I’m sure it will be awesome!

Quentin
Mail: quentindecarne@gmail.com
Twitter: Quentindc5

I’d be very grateful if you’d help to spread this post by emailing it to a friend, or sharing it in your social media streams. Thank you!