Smartup Time at TheVentury
At TheVentury, we believe that investing in the development of our employees is key to create value for our customers and the community. This is why we recently started to block four hours a week for every Venturer to do some research and gain new knowledge in fields where they want to improve, or in which they are interested. We love to learn new things and are proof that lifelong learning is not just an empty phrase. Since we try to give our clients as much knowledge about their project as we can, people are also highly encouraged to share their insights and takeaways from their research with their colleagues in form of small talks or workshops. In that way everybody benefits from the new gained knowledge of others, which also has a direct effect in improving our daily tasks and projects. This time our focus is: bot design.
Simon Morel on Conversational Interface Design
On Friday, we were lucky enough to hear from bot designer Simon Morel, partner & Head of Product at BotSupply some insights on how to make bots ‘that don’t suck’! The key takeaway is that when you are developing chatbots, you are designing an experience for the user, and you need to ensure that you’re adhering to design principles that have evolved in other spheres. Bots should be for everyone, and you need to design them appropriately.
Simon was keen to stress that bots are just a small part of AI-powered solutions, which will fundamentally change how we interact with computers – bots are, in many ways digital spokespeople for brands and organisations, and they should link to other AI-powered solutions within this role. This means that for bots, conversations are at the forefront of the way that users interact with the AI, so the focus on design needs to be on that, before you even get to the AI.
In the same way that we saw the transition from digital to mobile, we’re going to see the transition to AI, an insight also noted by Amir Shevat in Designing Bots. No industry is going to be unaffected – we won’t be able to imagine life without it. Simon’s perspective is that a chatbot is not just a bubble showing up on a website or within Facebook messenger, we don’t want to be limited. Current criticisms of bots tend to assume a simplistic view of bots, and don’t take into account that bots will soon use not just text, but also voice, eye movement, facial expressions, body positioning and hand gestures, and that AR and VR may soon have a big part to play, though Simon lamented that in Denmark, support for Danish voice conversations is lagging behind a bit!
You don’t need a computer to start designing a bot
Simon pointed out that when you’re designing a bot, you need to start with an analogue fashion (pencil and paper) – it’s faster and more accurate, and makes it easier to start working on the bot immediately. Don’t start with the copywriting or the character, since they will change rapidly and are no good as a foundation. Instead, start with conversational flows, and literally draw them as flow charts. Only once this has been finalised do we move onto the digital stage.
Apply the principles of character design in games and film
After this, you can design the character. Though sometimes we as designers don’t get to create the character at all, we should try to use what the medium of chatbots affords. Every interaction with the character should take the person into account that they are trying to imitate – what is the brand, what is the person, how do they sound? If this sounds familiar, it’s because these principles follows those of character design from games and film, and the lessons learned in that medium are just as applicable here.
The bot’s function follows the data and the design
Only after all this has been completed, Simon emphasized, we move to AI. AI must follow everything else, but it can’t be forgotten – if we don’t have any tech between the data and the design, it doesn’t matter how well-designed the concept is! Designers need to remember that AI isn’t magic, though. If we want a chatbot to be super-specific, then we need the data to back it up, otherwise we can never train the AI.
Finally, Simon closed by looking to the long term, and how to ensure that a bot doesn’t just die out as a flash in the pan. He returned to the idea of character design, and he questioned whether that should change as well? How do you do it without annoying the customers, and without people losing interest? Ideally, he stressed, you’d have 5-10 utterances of the chatbot’s phrases, and everything else should be subject to discussion with the client or designer. If users aren’t interested, that might be because of the service, but it might be because of how the character presents itself.
Since you read until here you are maybe interested in bots in general and writing outstanding characters for bots? Well, we are looking for a Character and dialog writer for chatbots to join our team!