211 34 Malmö
Sweden+46 735 124 email@example.com
Come along Kruso's applied technology journey in bot-land.
This is the first of 3 acts through which my experiences designing, building and launching a bot are narrated. This first act is about getting started, the second part will be focusing on the interaction design, and the third part is about development, testing and deploying. I have chosen an exploratory POV style – a form which will provide room for both babble and folly. So, the reader hoping for a 5-minute bot-building guide, I will recommend halt the reading here, and instead look for more results-oriented sources, such as: https://casual.pm/how-to/launch-chatbot. The curious (and patient) reader, however, will get a free ride on the passenger’s seat on my applied technology- maiden voyage into bot-land.
- Bot voyage!
Life as a digital consultant requires a supernatural physique. Or at least as long as one tries to maintain a self-image in which one is a digital front runner. The tech-race has no start, no finish line, no rules, and is constantly branching out in an endless number of sprints, initiated by new technologies emerging.
At the moment, I am out of breath struggling to gain a position in The Bot Race. I am constantly out of breath, but my FoMo provides me with sufficient energy to keep my feet moving. The flow of clever opinions about Bots and their enormous significance is infinite, and in my profession, it is anxiety provoking when you don’t have anything clever to add to the subject. At the Web Summit 2017 in Lisbon, I felt the pressure of the outside world for the first time. Bots seemed to have taken over conference entirely. In a spectacular fashion a full Centre Stage could watch and cheer as Sophia the Robot and Professor Einstein Robot, under the moderation of their "father", Dr. Ben Goertzel of Hanson Robotics, debated the implications of robots' involvement in human life.
And on the deliberately less entertaining conference track, “Talk Robot” one of Denmark's hot AI and chatbot profiles, Dennis Mortensen, shared insights from his project, X.ai. No matter what direction, I turned my gaze, it was met by a bot’s calculatingly friendly eyes.
Back in Copenhagen, the bot-noise was less piercing, and I successfully managed to escape my lack-of-bot related problems by engaging in client issues. For a little while. Until reality caught up with me at The Next Web conference in Amsterdam. Here, a trend talk claimed that “By 2020, the average person will have more conversations with their bot than with their spouse.” – A truly horrific prophecy; 2020 is right around the corner, and Kruso and I have yet to launch our first contribution to mankind’s next generation preferred conversation partner!! My key takeaway from the conference was, unsurprisingly, that NOW was the time for Kruso to make that bot, and that I had to be part of that pilot project! But. Alas. The next day at the office, I found myself overtaken by my Swedish colleagues who were announcing the test-launch of a chatbot they made for a big, juicy global Swedish company. Not even within my own company I was fast-moving. Sigh. Fortunately, the recognition of my defeat did not lead to apathy. Quite the opposite, actually. That's how you survive and thrive in the tech industry :)
With the air full of tech-love I started seeing bot-possibilities everywhere. There seemed to be no problem to which a chatbot was not the solution. Over and over again, I found myself in the role of a bot ambassador in project-related discussions, in which my bot-perspective had no relevance. Eventually, however, I realised that being a digital strategist, while at the same time spending the time and money of our clients to benefit my own hobby tech projects made no sense at all. So, I directed my search for an opening for my bot-pilot project inwards. And here it was: The new kruso.dk website. In theory, a match made in heaven project, with both great creative freedom and the interest of internal stakeholders.
I used our clever (and jumping-the-gun) Swedish team experiences as a starting point for a roadmap that would give my cure a quick and painless birth. The plan was based on the following main problem formulations:
Why should we use a bot?
How should the bot be experienced (Bot Experience)?
What should the bot be capable of, and in what interactions and scenarios should it be operating?
Which channels and technologies should we use?
What and how should we measure bot performance?
How do we work continuously to optimize the bot?
How can we translate our experiences into a specific product that can be offered to our customers?
I would encourage emerging bot architects to copy the above plan 100%, make their own critical considerations, and then, eventually, make their entirely own plan.
Although Kruso's own bot project is an internal one, it's by no means a hobby project, that is only implemented for the sake of the project itself. In our case, an obvious purpose with a bot on our public website is not to provide service to our customers, but a means of marketing Kruso and our technical skills. Thus, through an internal project, we will demonstrate the possibilities of using chatbot technologies, thus inspiring and facilitating individual conversations with our customers about how these technologies could support their respective client-facing conversations. Reformulated: Demonstrate the in-house skills by exploiting them in the concept and execution of the agency's own website. Specifically, the competence to be demonstrated is bot-building, and the method of demonstrating it is like a concept-layer on top of the usual, fully fact-oriented website, with the classic structure "who are we", "what can we do"," what have we made" and "contact", etc.
How should the bot be experienced?
Companies planning to launch a more service-oriented bot may benefit from experimenting with the interaction design first. But since Kruso's bot – at least in this first version – aims to give the user an experience, the next natural step is to determine the cure experience.
Basically, we have divided the bot experience into the following perspectives:
Placement / integration – where should it be experienced?
Visual appearance – how should it look?
Personality – How should it behave?
Placement / integration
In most cases, chatbots are implemented as a feature that is complementary to the main user experience and interaction concept of the solution that it is integrated into. In those scenarios the bot will live its own parallel life, as a kind of context for the main solution. A good example is Alka's chatbot, "Alma". The user starts his conversation with Alma by going to www.alka.dk and clicking on "Start Chat". This initiates the opening of a new window in which the chat takes place. For example, if you ask the question "How much does a car insurance cost?", Alma answers: "I cannot calculate a car insurance price, but you can go to Alka.dk and calculate or buy a car insurance. Click 'Calculate / Buy' (a link) ". The user should then click away from the chat to get the answer to his question. In essence, Alma is a navigation tool that allows the user to converse with content on the site, and the concept of placement and integration is thus "its own parallel track".
Here is a good opportunity for Kruso to show how to design and implement a chatbot, as a more integrated part of a comprehensive interaction concept. The vision for this solution is called "the call-based kruso.dk".
By design Sophia the Robot has a visual appearance that should to be appealing to humans. According to Ben Goertzel, the reason behind this design choice is to break down a “barrier of fear” of robots that humans have, allegedly. Only by bypassing that barrier, can we as human beings communicate in a relaxed and honest way with the robot, and thus it is a foundation for bot’s successfully getting genuine, non-distorted input from us. Let’s assume that this general rule can be applied in the design of a virtual bot. Of course, a virtual bot has a rather unintimidating appearance, as they are not physically present. On the other hand, the lack of physical existence may cause the bot to appear suspect. If we cannot see it, we cannot evaluate its intentions and why would we dare to entrust it our genuine thoughts? Here, the little avatar, which is often the only visual representation of the bot, comes into play. Design parameters for a chatbot avatar are:
Name - A real name? Or just CompanyBot, or ...?
Name / Style - Does the name have a special style, or should it just inherit from existing style sheet?
Icon size - How much inventory (space on the screen) should it occupy?
Motive – Should it look like a robot, a human being, an animal, or perhaps assume an abstract form?
Execution style - E.g., photorealistic style, line art, coloured illustration, etc.
Animation – Should the avatar have different variations, eg. for different situations?
Our design choices are (so far):
Name: Robinson the Bot
Name string / typography: Inherit existing CSS
Icon size: Multiple sizes, depending on scenario
Motive: Robinson Crusoe?
Execution Style: Iconic, Simple Illustration
Animation: Maybe. Ideas could be: I've answered, I do not have an answer, I've been waiting for your reply for a long time ...
Like the company name, Kruso's own chatbot is inspired by the story of Robinson Crusoe. Exactly what the person Robinson was like, will be solely for the author, Daniel Defoe, to provide an adequate answer to. Anything but will be a product of speculation, subjective perception and free imagination. And that's exactly what we're throwing ourselves into. We have luckily recently been through a rebranding process that resulted in a set of corporate values which can be inherited by Robinson the bot.
Boiled down to the essence these values are:
This set of values should be considered as a conceptual starting point, which will be tested and refined in the next act where we will work to defining, writing, and designing:
"The conversational interface"
Thus, Act 1 is concluded – stay updated!