There’s no denying that virtual agents – or “chatbots” (or simply, bots) – are experiencing a tremendous resurgence in interest, and along with that, a rapid advance in innovation and technology.
This resurgence is driven by four primary factors – it’s a perfect storm – a confluence of drivers causing massive interest, adoption, growth, and innovation.
The drivers are the growth in messaging apps, advances in technology, the opening of platforms, and commercial adoption of bots:
Figure 1 – The rise of bots.
Let’s explore each of these.
Driver 1: The Growth in Messaging
People love messaging. People use messaging apps more than they do social networks. The combined active users are larger in number than the combined active social network user base!1
Not only are people using messaging apps more than social networks, but messaging is also proving to be far more “sticky” than traditional apps.
According to The Economist, a quarter of all downloaded apps are abandoned after a single-use…with the exception of instant messaging. Over 2.5 billion people have at least one messaging app installed.3 This number is expected to climb to 3.6 billion within a few years according to Activate.
Not only are there more monthly active users on messaging apps than on social network, but people are also spending more time in messaging apps than social networks:
“People are now spending more time in messaging apps than in social media and that is a huge turning point. Messaging apps are the platforms of the future and bots will be how their users access all sorts of services. (Peter Rojas, Entrepreneur in Residence at Betaworks)”4
WhatsApp users alone average nearly 200 minutes each week using the service.
Driver 2: Advances in Speech, AI and NLP
As one may expect, there have been significant advances in the core technologies making up an intelligent chatbot.
Apple can largely be credited with making speech popular again, even if not improving much upon the speech technology of the time. In fact, Siri was initially powered by Nuance, a speech platform. 5 What Apple did do is make it ubiquitously accessible (at least to iPhone users) and viable. Speech was once the annoying interface to an IVR (phone tree), and now it became a somewhat usable personal assistant. It still had major shortcomings however especially in the context of the initial hype Apple placed behind it.
Furthering the excitement and adoption of speech technology was Amazon’s Alexa – a surprise hit, thanks in part to Amazon’s “undersell” approach – a contrast to Apple’s over the top marketing and subsequent letdown. While Apple, Microsoft, and Sony were fighting to be the “living room console”, Amazon discovered a new gateway in the home with Alexa – the voice gateway. Alexa has matured significantly, adding new skills as Amazon learns more about consumer vocabulary and behavior. Alexa is making a big push in the home automation space, and companies like Ford, LG and more have all unveiled plans to have Alexa integration.6
Google, fearful of being left behind, is rapidly playing catch-up in this space. They have released their own version of Alexa hardware…Google Home. In addition, buying up companies in the AI space, such as API.AI. While Amazon has first movers’ advantage for the moment, Google is in a powerful position of knowing everything about you – from your calendar to your flight schedules and your interests. A potentially powerful combination.
Microsoft too is getting into the game, partnering with Harmon to add a physical home device (ie. Speaker) to their Cortana virtual assistant product7.
Of course, it’s not just about the speech recognition or the in-home device. Tremendous advances have also been made with NLP (Natural Language Processing) and AI. AI and machine learning now drive our cars, manage our health, our lives, automated trading, and guided missiles. The advance in AI is of such an extent that that we now even have AI ethics conferences. An alarming new term has been coined: an “intelligence explosion” – referring to the exponential leap in cognitive ability that notables like Elon Musk and Stephen Hawking fear could one day spell doom for the human race!8 This closed-door conference resulted in a letter pleading AI researchers to consider the safety and ethical implications of AI. Many of the leading AI researchers have signed off on the letter…in fact, more than 8000 leading researchers and scientists so far.
The purpose of this paper is not to instill fear, and ‘doomsday’ is not around the corner. What these facts do however illustrate is how significant the advances in technology have been, driving the adoption of bots for the everyday user and consumer.
Stanford University host a project (and panel) entitled the “One Hundred Year Study on Artificial Intelligence (AI100)”9.
The study points to immediate gains in eight key domains – for example, autonomous transportation will ultimately result in new urban organizations as people need fewer cars, and in healthcare, AI-based health outcomes can improve the quality of life for millions10.
As for what’s next in AI research, the report states that “the field of AI is shifting toward building intelligent systems that can collaborate effectively with people, including creative ways to develop interactive and scalable ways for people to teach robots”. Individual areas of improvement and research include improving natural language processing to interact via dialog and not just react to formatted requests. Neuromorphic computing seeks to mimic biological neural networks to improve hardware efficiency.
The race for AI resources is also intense. Underscoring the strong push for these resources, Google acquired DeepMind, a hot AI startup, paying $400 million for a 50-person company. Suddenly Google owned the largest available talent pool of deep learning experts in the world.11
The race is underway. One can hope that the consumer is the ultimate beneficiary. In the short term, we will have to deal with a fragmented ecosystem (VHS vs. Betamax anyone?) and hope that providers interfacing with speech, voice, and AI systems add support for multiple technologies.
Driver 3: Opening of Platforms
Much like the ability for the iPhone to have a community of developers building Apps, leading messaging platforms are opening up their platforms in the hope of building a thriving ecosystem on their technology.
The Economist talks about the birth of the bot-economy, largely crediting the Russian originated company12, Telegram with being the first to open its service to their 100m users, launching a bot platform and a “bot store”13. This allows developers to build new solutions and products utilizing the core bot IP.
During the last 2016 F8 conference, Facebook announced the opening of its Messenger APIs to third-party developers, desiring to create their own bot-economy and build a strong partnership with the developer community to come up with new solutions while increasing Messenger penetration.
At the time of writing, WhatsApp stubbornly clings to a closed paradigm – however, with the acquisition of WhatsApp by Facebook, we can expect either a clear line of demarcation (Messenger being used as the bot platform, WhatsApp as the human-to-human platform), or perhaps an eventual consolidation into a single ubiquitous messaging platform.
Google, of course, will not stand idly by. They acquired API.AI, a technology that allows you to build your own Bot. Google also open-sourced its TensorFlow library for machine intelligence14. As may be expected, Amazon not only opened up the Alexa platform to build new Alexa skills, they’ve also made available the voice services to allow developers to build new solutions incorporating voice and speech15.
All of these moves will ensure new products and solutions coming into the market place, built on these technologies. Just as Apple made “There’s an app for that” popular, soon there will be a “bot for that”.
To politely steal, but perpetuate, The Economist’s term – the “bot economy” is here.
Driver 4: Bot Success in a Commercial Setting
Moving from the theoretical future of AI and machine learning and into the more pragmatic “now”, we’re seeing tremendous adoption of Bots in a commercial context, both for customer care/service and sales. In Jacada’s world, we refer to Bots in a customer service context as Virtual Intelligent Agent Assistants.
Gartner estimates that the current VCA market, as measured in revenue by the top 18 VCA vendors, to be at around $450 million16, with an expectation of a 25-35% YOY growth. Grand View Research believes that by 2024, this market will be in excess of $12 billion. In the same report, Gartner state that they believe by 2020, 25% of customer service and support operations will integrate virtual customer service robotic technology in their engagement channels, up from less than 2% in 2015. More optimistically, at least one VCA vendor makes the claim that by 2020, 85% of customers will manage their relationship with an enterprise without interacting with a human.
Regardless of “who’s right”, there can be no denying the tremendous uptake in the adoption of VCA technology.
Part of the appeal of VCAs from a corporate point of view no doubt is cost. A standard inbound voice call costs an industry average of around $12 per call, compared to $1 for a VCA chat session. Compounded over millions of interactions, this results in significant cost-saving potential.
As customers rush to not only drive down costs but to also improve the customer experience, we can expect tremendous growth in the VCA space.
What Makes Up a Modern Virtual Customer Assistant?
There are a number of components that modern virtual assistants will need to provide. We depict these in what we term the Periodic Table of Virtual Agent Elements:
Figure 2- Jacada’s Periodic Table of Virtual Agent Elements
Let’s briefly explore these elements:
- Dialog – Humans want interactions, not single request/response conversations where each subsequent request has no continuity to the previous response.
- NLP – A core requirement for any capable VCA is the ability to understand all permutations of a customer request.
- AI – Without adopting AI and machine learning, your VCA will never improve unattended. A Rules driven approach is an acceptable short-term approach to lower implementation time and cost, but a mature deployment needs to embrace machine learning.
- Transactional – Many VCAs today are limited to information retrieval and display. A true VCA should be transactional, meaning it has the ability to fulfill an entire self-service task, such as filing a claim or updating information.
- Analytics – VCAs should be deployed with continuous improvement in mind. This means understanding problematic areas and identifying areas of improvement which requires analytics.
- Context DB – VCAs should become an extension of your assisted service workforce in as much as they need to have the same contextual capabilities as your voice agents have today. The VCA should know about your previous interactions in order to drive successful outcomes.
- Speech – VCAs should not be limited to text input or output and should support multiple modes of operation.
- Search, FAQ and KM – VCAs fundamentally process information and either need their own internal KM with appropriate taxonomy, or the ability to conducted federated content searches across the diversity found in an enterprise.
- Multi-channel – VCAs should not be limited to just your website and should be offered at multiple points of engagement, to both digital and voice callers.
- Connectivity – Inevitably a portion of VCA interactions will still require human assistance. These should be escalated seamlessly and with context.
- Open – With the rise of 3rd party messaging apps, VCAs should learn to live within a parent container.
- Personalization – It is no longer acceptable for a VCA to use a “one size fits all” model. Instead, each interaction should be personalized based on the customer, his/her context, and interaction history.
- Flow and intent-driven UI – A VCA doesn’t necessarily imply just text chat (in addition to speech). It can be flow driven where the user is presented the most likely set of options or intent-driven where the customer more explicitly describes their intent.
VCAs certainly are not limited to a support role. Conversational Commerce is a term coined by Chris Messina from Uber referring to the intersection of messaging apps and shopping17.
Facebook worked with the leading payment providers PayPal, Braintree and Stripe and even poached PayPal president David Marcus to run its Messenger division, underscoring the value Facebook is placing in “conversational commerce”18.
To send money to someone within the Facebook messenger, a user simply:
- Opens a chat with the person they want to send money to
- Click and enter the amount they want to send
- Click Next to add your debit card and then click Pay
Figure 3 – Facebook messenger payments (image via TechCrunch)
Let’s be clear – showing an example of payments is to some extent missing the true meaning of conversational commerce. Instead, think how sophisticated bots will allow the customer to interact with the organization to get what they want, without needing to learn some unwieldy complex web or app interface.
Conversational commerce lets you take a different approach.
It allows a user to access all the required functionality through the familiar chat interface.
For example, WeChat in Asia, boasting ~700m monthly active users, already allows people to make payments, e-commerce purchases, hail taxis, order food, customize and order a pair of Nikes, host a conference call and more all through a chat interface19. Imagine how complex this app would be with a traditional interface. Chat solved the complexity.
This is the power of conversational commerce.
The “bot economy” and “conversational commerce” are terms you need to be familiar with. Your organization needs a strategy of how to embrace these concepts. With the rapid rise in messenger adoption, the rise of the millennials, and the advances in technology, the bot or VCA is going to be “must-have” technology.
In closing, we leave you with interaction from arguably the most infamous computer interaction of all time. In Stanley Kubrick’s classic movie from 1968, “2001: a space odyssey”, written by him and renowned sci-fi author Arthur C. Clarke, there is a great dialog between the main (human) character, Dave, and HAL 9000, the AI Computer. Dave and his colleague Frank are considering disconnecting HAL due to what they perceive as a malfunction. Attempting to conceal their conversation, they are unaware that HAL can read their lips. The conversation goes as follows:
|Dave Bowman: Open the pod bay doors, HAL.|
|HAL: I’m sorry, Dave. I’m afraid I can’t do that.|
|Dave Bowman: What’s the problem?|
|HAL: I think you know what the problem is just as well as I do.|
|Dave Bowman: What are you talking about, HAL?|
|HAL: This mission is too important for me to allow you to jeopardize it.|
|Dave Bowman: I don’t know what you’re talking about, HAL.|
|HAL: I know that you and Frank were planning to disconnect me, and I’m afraid that’s something I cannot allow to happen.|
|Dave Bowman: [feigning ignorance] Where the hell did you get that idea, HAL?|
|HAL: Dave, although you took very thorough precautions in the pod against my hearing you, I could see your lips move.|
|Dave Bowman: Alright, HAL. I’ll go in through the emergency airlock.|
|HAL: Without your space helmet, Dave? You’re going to find that rather difficult.|
|Dave Bowman: HAL, I won’t argue with you anymore! Open the doors!|
|HAL: Dave, this conversation can serve no purpose anymore. Goodbye.|
And in the continued spirit of HAL, this article can serve no additional purpose anymore. Goodbye.
[About the author] As Chief Marketing Officer, Chris’s responsibilities at Jacada include global Go-To-Market strategy, corporate positioning, and marketing strategy and campaigns. Chris has over 20 years of experience in product management, marketing and software development having held several senior leadership positions. Before joining Jacada, Chris founded a successful software consulting company providing large scale software systems to Fortune 500 companies. Outside of work, Chris enjoys app development, automated trading algorithms, and provides pro-bono legal services. Chris holds a BS Computer Science and a JD and is admitted to practice in the State Bar of California
1 http://www.businessinsider.com/the-messaging-app-report-2015-11 2 http://www.businessinsider.com/the-messaging-app-report-2015-11 (chart) 3 http://www.economist.com/news/business-and-finance/21696477-market-apps-maturing-now-one-text-based-services-or-chatbots-looks-poised 4 http://beaconing.eu/2016/09/27/the-chatbot-era-has-begun/ 5 https://en.wikipedia.org/wiki/Siri 6 http://www.businessinsider.com/amazon-echo-success-could-spell-big-trouble-for-google-2017-1 7 http://www.theverge.com/2016/12/14/13951526/microsoft-cortana-harman-kardon-speaker-amazon-echo 8 https://www.wired.com/2015/01/ai-arrived-really-worries-worlds-brightest-minds/ 9 https://ai100.stanford.edu/2016-report 10 https://ai100.stanford.edu/sites/default/files/ai100report10032016fnl_singles.pdf 11 supra 12 https://en.wikipedia.org/wiki/Telegram_(software) 13 http://www.economist.com/news/business-and-finance/21696477-market-apps-maturing-now-one-text-based-services-or-chatbots-looks-poised 14 https://www.tensorflow.org/ 15 https://developer.amazon.com/alexa 16 Market Guide for Virtual Customer Assistants, Gartner, 23 November 2016 17 https://medium.com/chris-messina/2016-will-be-the-year-of-conversational-commerce-1586e85e3991#.sx60xirsw 18 https://techcrunch.com/2015/03/17/facebook-pay/ 19 https://chatbotsmagazine.com/11-examples-of-conversational-commerce-57bb8783d332#.nqaojae28