CHAPTER 3

Use Cases of the Best Practices and Worse Case Scenarios

Defining a Cognitive Interface by What It Is Not; the “Japanese Hologram Girlfriend”

I created this case study for a large audience of MBAs and staff at Cambridge Judge Business School in autumn semester 2017. I won’t repeat the case study in this book because it is freely available online as a podcast on YouTube accompanied by the visuals in a SWAY PowerPoint.1 The essence of the analysis was that the interactive hologram created by Gatebox in Japan was not a cognitive interface. It does not grow or learn. It is an excellent, high-grade animated 3D avatar.

The Hologram Girlfriend ever deployed in a small semitransparent cylinder for B2C use or mass consumption, to date largely in Japanese though there is uptake for her English version in the United States. However, the personality contained in the box is not an AI bot hologram as it is on a loop or repeated content. It doesn’t learn from its owner or from interactions with humans to expand its repertoire.

Having said that, its performance as a standalone 3D hologram satisfying the needs of a particular market segment is remarkable. Its best innovation, rewarded with a fairly quick M&A takeover by a larger corporation in Tokyo, is the Hologram Girlfriend’s success in being an IoT device in home environments, connected to the owners’ appliances and being able to text him via his smartphone, a NLP chatbot in 3D communicating as a 2D replica with her owner.

It may be the subject of another book to look at chatbot history by continent, for example, in Africa, Asia, India, Australia,2 the Americas.3 We have had RFPs from South America and we noted that the most popular instant messaging app there is Telegram, which is also running thousands of interactive, basic NLP chatbots for a variety of use cases.

I stumble across Japanese, Korean, Chinese, Indonesian, Malaysian, and Singaporean chatbots constantly, not to mention the thousands coming on the Indian market. Many have been around for decades, like these corporate use cases for 2D chatbot applications via instant messaging platforms by a South Korean AI company that was established in the early 1990s.4

Eight Use Cases for Industry Verticals

For our marketing collateral, my company AI BaaS defined the following use cases:

1. Wayfinder and infotainment at malls, airports, train stations, and sports stadiums including sponsored marketing targeting “want to know” users during their chat sessions with velmai AI bot holograms on site in 3D multimedia format.
2. Conferences and trade shows—see the Portia experiment. Saves time, money, and energy when you can send a bespoke 3D hologram avatar instead of staff! She made a speech to AI experts at AMLD and took questions from the floor. The hologram can then be installed at the event for further one-on-one interactions.
3. Tourist guide and hotel concierge—similar to use case 1 but more bespoke content answering the FAQs of hotel guests and destination marketing goals.
4. University campuses for security and timetabling queries regarding room changes in real time. 3D is a mascot of the university that we can grant student teams access to, that is, they can upload new info via a CRM. The 2D replica can disseminate this via the university’s and student clubs’ social media pages. We have also been asked to do a Student Welfare bot for psychological support on campus.
5. Government departments and NGOs: again as Infotainment Wayfinder on site in large government buildings and spaces. 2D bots for B2C interactions and customizations to bring government services to life in real-time conversations to get bureaucracy done.
6. Point-of-sale sales assistants in store or for special promotional events—the 2D replicas on social media and the corporate websites reinforce the sales work and interaction of the 3D hologram AI bots live on site.
7. Marketing avatars, for example at banks, insurance companies, and other services that have cut back on sales and reception staff. But as the first point of contact for clients, business partners, and suppliers, a 3D AI bot hologram on site is a 24/7 ambassador that can bring the IoT of the firm and the essential info together in “one brain.”
8. Education generally: we have had many RFPs for tutor bots, for example, at MOOCs.

RFP Case Studies for “Dos & Don’ts”: Hypotheticals Based on True Events

Case Study #5: Diplomatic Mission’s Facebook Page and Messenger 2D Chatbot

This deal came about after a call from or to the Deputy Ambassador of an Embassy in London (I can’t recall how this RFP was initiated exactly). They wanted us to solve a political, technical, and social problem. Curiously, it was to do with Brexit and populism.

Without giving too many details, we can say that this entity ran a Facebook page that had thousands of fans visiting it regularly. At the time we drew up our proposal, there were 4 or 5,000 fans or followers on the FB page). Sadly, that number included numerous trolls and a few very aggressive Brexiteers who saw a conspiracy in everything.

So our task was to build a Facebook Messenger bot tailored to the mission statement and purpose of the page. It was to meet and greet the visitors in real time, for example, “Hi Thanks for visiting our FB page. How are you? Can I help you with information?” And then a tree structure with “pick your answers” that would graduate to a free form or unstructured response of the bespoke bot replying “naturally” to the spontaneous comment or question.

The main task, however, was for the bot to solve the problem of constant, random trolling of the site. The challenge was for it to identify abusive language and give automated warnings 24/7. The idea was to take the heat out of the escalating “discussions” among the “fans” of this FB page by having a calming, neutral, and “diplomatic” bot take care of basic communications without being provoked, angered, or upset by any personal attacks or verbal abuse by the visiting, “hostile” Facebook users.

It would have been the perfect case study to show how even a basic messenger bot could defuse hate speech online especially on Facebook, which is notoriously slow to remove offensive comments. Despite the platform making billions in profit a year in every corner of the globe, Facebook remains unable to deal with or police the nasty perpetrators of bullying and trolling.

Again, almost de rigor, we were about to start building this bot after it got sign off from all the stakeholders. Then Brexit took a bad turn and it was evident that those involved would have to change their policy and wind up this Facebook page as an “outreach” or PR project in the UK. The client could not justify the budget of £10,000 for the final 12 months of this online campaign before they were obliged to wrap it up, due to political changes and social pressures regarding the viability of running the project amid the furor of populist beliefs, unproductive negotiations, and the British public’s escalating anger or frustration, as is the case now on the October 5, 2019, a matter of weeks before the Brexit Deal—or not.

Figure 3.1 © Cliff Lee, CCO from velmai Ltd. “Albert the Online Butler” (an iteration of Batman’s butler Alfred) was the prototype that attracted the attention of this client in London. Albert was also voted the best innovation at the London Digital Travel Summit in 2017. https://youtube.com/watch?v=6coVEQzE27A

Our chatbot company had invested time in face-to-face meetings and PowerPoints tailored to the goals and objectives of the job. The 10K was not the decisive factor for us, obviously a first mover client of this stature would have been a boon beyond the initial revenues! I later turned this proposal or request for a proposal (RFP) into a live case study for my Masters of Science students at an international business school in the south of France.

The challenge was to revive the project with other branches of this diplomatic envoy and organization in other countries. The course was on “Negotiations and the Use of CRMs.” It turned out to be a very productive use of the students’ imaginative ability to put themselves in the role of the entrepreneur and the nongovernment or NGO decision makers. They nearly closed two deals in Germany and Italy, with several other pitches drawing positive feedback and scoring a won deal by the end of semester, which they dutifully tracked in their ERPs in the cloud.

Case Study #6: International TV Station and Digital Broadcaster: Online Bot as Infotainment

This deal was won in a strangely “covert” way, without pitching directly for it. Instead we were unwittingly put into a “live experiment” and then chosen from our peers! We had received an invitation from New York for a one-day closed door seminar on chatbots with leaders from our competitors and our peers. The number of actual BaaS providers was limited to just two dozen at the most.

I didnʼt understand the conference organizers’ business model at the time. As a one-day Londoner seminar for 20 of my competitors and less than a roomful of attendees, there were also no obvious sponsors of the event. Business participants attended for free, on the entrepreneur side. As we later realized, the people in the room, apart from the BBC, which interviewed me and two others on our AI bot panel in the afternoon, were from big corporations. They were purchasers or buyers.

Clearly, they were paying to “go undercover” to see us perform in our natural environment as was later explained to me by several HR experts. They had seen parallel practices in their fields obviously. Some of my peers unwittingly discussed big flops of bot deployments and failures in delivering, believing we were all there as BaaS providers.

The attendees did not have name tags and I only found out where some of them were from over lunch. I luckily landed on the table with the TV corporation that later gave me a call. It was between us and another company regarding winning the purchasing contract!

We were asked to quote on:

• A 2D bespoke bot.
• It was to do be a type of menu bot that helped users navigate the content of their online sites.
• It was the beginning of on-demand service in the UK and this corporation was already at the forefront for certain verticals.

The Pitch and Outcomes

In response to the stated needs of the client outlined mentioned earlier:

• We created a name and character for the avatar which his team liked.
• However, the marketing director was afraid of the bot's automation and typically for marketing and business development (biz dev) people, thought there was an in-house plot afoot to undermine their authority and control of the site.
• The content had millions of users globally and we agreed to limit a first deployment to Europe only.

This was suddenly called off, after we had gone to the effort to send them a bespoke licensing agreement. The TV company did not replace us with a competitor bot. To my knowledge, there is still no botification of this site or any of the content available online.

The only clue we had to yet another cancellation preproduction was that the Project Manager who had engaged with us over lunch, actually a senior executive, had been seconded suddenly to another country and there was “new management” in the UK subsidiary. They were not open to continuing discussions of bots and botification to support users to find content on their site.

It could have helped users to locate videos and films they wanted to view. More importantly, it could have promoted and publicized content they were not going to discover on their own. Being an in-house trusted corporate publicist, on-demand 24/7 in any language would have sent hit rates and user engagement skyrocketing.

Figure 3.2 © Cliff Lee in Devon. “Charlie the Spoof Newsreader” was a prototype 2D chatbot for this content managing bot that was to promote the shows and topical documentaries online and also conduct “mini polls” and user quizzes on behalf of the brand or client

Yet again, internal politics and problems at a global multinational corporation led to the obstruction of low-cost innovation and the delay of this emerging tech adoption.

Case Study #7: An Internationally Franchised TV Show: 2D Search and 3D AI Bots

We won the invitation to pitch for this TV deal, similar to the other on demand one discussed in this section, in an unexpected way. I had given a guest lecture at the prestigious Cambridge Judge Business School. It was announced as a talk on AI bots from an industry perspective. And the next thing I knew there was a huge queue to get into their biggest auditorium, with students having to sit in the stairwell and stand in the aisles.

After I spoke for nearly an hour on the Japanese Girlfriend and botification generally (see my discussion of this in Chapter 1), someone who had RSVPed but could not attend in person met with me one-on-one in Cambridge. They worked at a famous broadcaster and within a month or two, we were having a meeting with the Executive Producers of this nationally well-known TV show, which was also available online.

They had spent a lot of money on Amazon Alexa skills.5 And they wanted us to fix them because they weren’t really working to their satisfaction. The growing customer, user, and developer dissatisfaction with the product and bots as a service is relatively easy to find in (ironically) critical reviews on Amazon.

Review of an Amazon Echo book on how build Alexa Skills.6 I discuss the problems arising socially and psychologically with the mass uptake of Amazon Alexa globally in Chapter 6, in the conclusion and afterword of this book.

The skills that had been taught to the Alexa platform were meant to give search results from their massive archive via voice queries. However, the Amazon API and bot development was having trouble:

1. Understanding the voice queries;
2. Responding appropriately; and
3. Finding the correct results.

One of the biggest problems was that to create the skills and then correct these issues was taking a lot of time, more human resources than they had anticipated and wasting a lot of time of their TV production crew and back office. That was not what they had bargained for when paying for this development. And then paying more again to fix the problem.

Why were we offered the job and then it was rescinded within a week? Because they didn’t want to pay us because we were a no-name startup. We were meant to fix Amazon Alexa skills for this famous TV show forfree and then go on to provide our better-performing AI bot with full integration into their archive and database of search results/user demanded content for free. Of course we said no. No, we didn’t want the fame of working without pay for a famous TV production. Global or not. No thank you.

Case Study #8: Chamber of Commerce 2D Chatbot for Membership Recruitment

Again I am not sure which side initiated this first. I think we were interested in becoming members of this chamber. Then the CEO became personally interested in our technology, pilots, and innovation. We became his protege project within the organization for at least 12 months leading up to the U.S. elections at the end of 2017.

Figure 3.3 © Cliff Lee. “Sir Loin Your Virtual Butcher” and several of our 2D chatbot pilots inspired the prospective client to ask us to “fix Amazon Alexa” because the food-related “Skills” were not working. See Sir Loin’s avatar in my Introduction which discusses this “alternative farm butchery and sustainability comms” case study

Figure 3.4 © Cliff Lee, cofounder of viledge Ltd. This was the prototype startup before velmai. It was a social forum before social media i.e. before Twitter, chat apps and Facebook. Its purpose was remarkably similar: to connect people with each other through the use of “undercover” or incognito 2D chatbots. We were pioneers in deploying online bots who texted each other and the human members on the site, recruiting user numbers into the tens of thousands all on their own, without disclosing themselves as chatbots

In one way, this was an incredibly simple bot to build, and in another, it was hugely challenging in terms of stakeholder interests! It took over a year to negotiate due to us being put in a role of service provider with special privileges. For example, we were going to be used as an exemplary proponent of chatbots in business. This would lead to us running a forum and networking group within the organization for members to explore notification, not just with us obviously but with all service providers.

We were about to create this female avatar who was to be the websiteʼs meet and greet bespoke bot. Her primary purpose was to recruit new members and also keep existing members abreast of events and initiatives. What happened? Without giving away the name of the longstanding entity, I can only say that a monumental election took place in their country of origin. From one month to the next, the CEO was replaced. Most of his team was sacked. And we were suddenly notified that the whole deal was off.

Figure 3.5 © Realfiction, Copenhagen, 2019. Image of a hardware device to house AI bots in hologram form or high quality holographic visuals. In 2020, we see the beginning of a shift from 2D online platforms to Mixed Reality devices that enable a person to obtain and hear information via voice. The hologram design is improving rapidly so that they will be able to replace website interfaces. People will interact with animated images, moving pictures and command the 3D visuals by speaking, interacting with gestures and chatting. What is already known as Conversational Commerce (applied to business) or Conversational AI

---