Why CAI is the Future of All Business
This story is a rough-draft. Check back later for the fully-polished story.
The missing link to widespread adoption of hyper-automation is a scalable interface.
Efficiency
The essence of all successful business is efficiency.
Let me take a step or two back. The goal of all business is profit: generating more revenue than the cost it takes to generate that revenue. It also should generate some value for some customer that is greater than the customer’s perceived cost of that value. The more profit for the business, the more success for the business. The more value per cost for the business’ customer, the more success for both customer and business.
Profit means efficiency. Efficiency comes in more shapes and sizes than I could ever list because cost and value both come in more shapes and sizes than I could list. Still, this one abstract concept defines all business success: efficiency. The company that can make a widget while spending half the resources compared to their competitors making the same widget will be more successful in the long run.
One of the biggest expenses, if not the biggest expense, any business must deal with is the cost of human time. What is the one clear way to do any business process more efficiently? If that business process is automatable, then clearly the way to become efficient at using that process is to automate it. We have seen automation grow in sophistication throughout all of human history. Starting with water and wind powered mills, and eventually using a mobile form of power called steam, the industrial evolution was a global-scale disruptive technology founded entirely on automation-based efficiency.
In our recent times, what technology is making business processes that were once hard to automate more automatable?
AI.
Conversational UI is likely the next big UI development in the software industry since the mobile app. We’ve seen the evolution in UI in the past from command-line desktop applications, to graphical user interface (GUI) based desktop applications, to web applications, to mobile apps. The next way that companies will interact with their customers and employees will be via conversational UI. It’s a hot topic now in many industries. Many companies are reporting significant returns on investment.
But users expect chatbots to be smart. The first interesting thing to note about conversational UI is that it is so natural for users to interact with. In theory, there is nothing a native speaker needs to learn in order to interact with a conversational agent, which is very different from older UI techniques. The second interesting thing is that, because natural language is so natural, users unconsciously expect more from a conversational agent than they do from the previous forms of UI. There are a number of additional expectations that users are already imposing on CAI-based applications. They expect the artificial agent to act like a human. Consider all the work reported online that is now associating computer vision capabilities with conversational AI even though vision is not at all necessary for CAI. Users expect more from a conversational agent than they do from any other UI in the past. They expect the agent to remember what they told it a few seconds ago. As surprising as it may seem, remembering context is not a common chatbot feature yet. There are a few other challenges listed below.
Developers also expect chatbots to be smarter. If these challenges can be addressed, even within certain well-defined domains, industries or use cases, the potential for CAI automation is amazing. The natural way that CUI leads so directly and so quickly into CAI has motivated some of us, including the Rasa founders, to believe that the eventual evolution of CAI will be completely autonomous businesses, even within the insurance space. https://blog.rasa.com/conversational-ai-your-guide-to-five-levels-of-ai-assistants-in-enterprise/
Conversational artificial intelligence (CAI) is the future of business. I predict that a majority of businesses will have a majority of their customer interactions taking place via one or more virtual assistants (bots) within my working life. This is because:
(1) Customers want to interact with a business as a single, unified, intelligent entity, not with isolated employees who don’t remember them and don’t have a complete picture of their own company,
(2) If CAI can be built to work right, written and spoken natural language is the most natural and efficient mode of communicating for people,
(3) CAI is almost good enough already,
(4) Business’ customers are in fact people,
(5) Many other threads of business automation are already becoming mature enough to naturally support that interface, and
(6) Conversational UI is the one UI style that can transfer consistently to any device and any form-factor to produce a unified persona for almost any business.
Customers may call them “virtual assistants” or something else instead of CAI. However, I focus on CAI in this publication because (1) it is the linchpin that will hold all of this together and (2) it is central to my own lifelong technical interests.
CAI (pronounced “kai” and rhyming with “hi”) is the best term I have to encompass a great new world of business automation based on AI and NLP. It may start with a simple conversational interface to a small application, like a rule-based pizza-ordering chatbot. In the end, I predict that that it will expand well beyond that for many notable businesses. There are two reasons — two forces that will drive the expansion of CAI. First, the user’s expectations of human-level conversational skills will incentivize CAI developers to expand the artificial intelligence capabilities of their agents beyond NLP because there is no other way to provide a convincing dialog without such things as efficient knowledge representation and knowledge engineering. Second, the myriad threads of business automation that have been growing separately and independently for years will naturally come together to support that need for intelligence; they will most naturally become unified behind the business’ most ubiquitous interface with its customer: CAI.
People naturally want to interact with businesses, large and small, as a single intelligent entity. How many times have you been frustrated with a big business when you call their customer support mob a second time and find that they have no memory of your first call? Sure, you are talking to a different person each time — and that’s the point. Why would you want to talk to a different person each time you interact with a business? I believe that, by far, the most commonly preferred customer experience is one in which there is a single entity the customer talks to and that entity never loses any ground in advancing your request.
Better yet, how often have you needed to call a business and you could not because it was not during “business hours”?
There is a trend toward more self-service. Look at travel, home buying and selling, etc.
…
The Naturalness of CAI
Including no need to learn a new interface for the old and the young.
The Adequacy of Current CAI
x
Markets are People too
People like to type and talk. People aren’t born knowing how to use a graphical user interface or programming/query languages. Even when people learn these things, these things change and they have to learn them all over again.
Customers expect a frictionless, interconnected shopping experience. — “One Home Depot”, 12/11/2019
People think of a business as a single, unified, coherent entity. Do they always act that way? Or have you noticed any businesses that sometimes act schizophrenic?
Businesses need to send a consistent message to their customers through marketing and sales channels. The one way to guarantee that is to drive both communication channels from the same knowledge base. This becomes so much harder to do with humans as the substrate on which those channels are built when those channels are learning from and responding to customer information independently without informing each other.
A Unified Business Persona
Marketers often talk about a customer personas. But how does your customer perceive you, as a business? Does the customer think of you as a single, intelligent entity? Or does the customer know that you are a collection of employees and departments with competing personal agendas and an inability to work together? Getting your business persona right can be life-changing for your business.
Customers of every business, both small and large businesses, need a smart, seamless, and scalable persona who can help them at any point along their customer journey, in any channel they happen to be in, one persona that bridges the divides across a business’ disjoint set of orgs. It’s easy for small businesses to provide this. It is very, very hard for large businesses to do the same.
McKinsey & Company (Nicolas Maechler, Kevin Neher, and Robert Park) wrote a great explanation of why businesses need to focus on end-to-end customer journeys instead of individual touchpoints.
Businesses need to present a single, coherent persona to the user. The customer shouldn’t need to figure out the right employee or department to talk to, let alone figure out how to contact that employee or department in any way distinct from any other employee or department. Someone who has flown on an airline who lost their bags knows this is true. It’s always some other coworker’s fault. There’s always some other number to call to get an answer about where your bags are.
Conversational AI is the key to ensuring big businesses can present a seamless, painless customer journey. Not only is an intelligent chatbot able to be the single persona that a customer talks to in any situation, on any channel, and at any point along their journey, but the work needed in brining all business departments to bear at the most appropriate times in those customer journeys will teach that business how to become unified and productive for the user’s benefit in whatever conversation the user wants to have. No longer can that business hide behind a fragmented user interface, with one department being responsible for one touchpoint and another department another touchpoint. It will eliminate tribalisms and politics, or that business will fail. Those businesses that succeed will give their users something truly magical. And in return, those users will make that business dominant in its industry.
The Unifying Potential of CAI
There are already forces working on businesses to transform their separate databases into a unified knowledge graph. It’s simply that knowledge graphs are so useful for so many things, and for business to become more automated, its data must be structured in a way that is useful for all of its processes.
Graphs “are a prerequisite for achieving smart, semantic AI-powered applications that can help you discover facts from your content, data and organizational knowledge which would otherwise go unnoticed,” says Atkin. They help corporates organize the information from disparate data sources to facilitate intelligent search. They make data understandable in business terms rather than in symbols only understood by a handful of specialized personnel. And they speed digital transformation by delivering a “digital twin” of a company that encompasses all data points as well as the relationships between data elements. “By fundamentally understanding the way all data relates throughout the organization, graphs offer an added dimension of context which informs everything from initial data discovery to flexible analytics,” he says. “Graphs give corporates the ability to ask business questions and get business answers and value. These developments promise to enhance productivity and usher in a new era of business opportunity.”
The value of unified data when trying to intelligently automate different kinds of tasks is no where more apparent when trying to automate the communication between that business as a single, seamless entity and any one of its customers.
Cognizant predicts that conversational AI will be more preferred by customers as an interface to businesses and that it will be more contextual and personalized. Cognizant also predicts that there will be a need to integrate with back office systems and software and data and also a big need for a lot of training data. I believe that the integration part is very true and will require something like a sophisticated Knowledge Graph or ontology.
Curated Knowledge
Streaming Data
In Every Company is Becoming a Software Company (or Every Company is Becoming Software), Jay Kreps makes two important arguments. The first is the same as this story’s main point: businesses are becoming defined almost exclusively as a fully-automated software-driven set of processes:
I believe that there is an accompanying change — one that is easy to miss, but equally essential. It isn’t just that businesses use more software, but that, increasingly, a business is defined in software. That is, the core processes a business executes — from how it produces a product, to how it interacts with customers, to how it delivers services — are increasingly specified, monitored, and executed in software. This is already true of most Silicon Valley tech companies, but it is a transition that is spreading to all kinds of companies, regardless of the product or service they provide.
…
Of course, this is not to imply that companies will become only software (there are still plenty of people in even the most software-centric companies), just that the full scope of the business is captured in an integrated software defined process.
Jay’s second point is about streaming data. More on that later. …
CAI as the Linchpin
A conversational UI will become the point where the unification and automation of disparate business functions becomes essential.
CAI Won’t Come from Nowhere
What business processes and applications will likely evolve into, or be replaced by, CAI, in whole or in part?
CAI will Change how we do Business
What job functions will be replaced or supplemented by CAI:
- Middle Management: Robot for a Boss? Most Workers Wouldn’t Mind.
CAI will Impact the Labor Market
These days when AI is over-hyped and often feared, CAI one typical instance of this. What are we to think about the dark predictions of AI taking our jobs. In The Economics of AI Today (published 18 Jan 2020), Juan Mateos-Garcia reviews the latest The Economics of AI conference. Their perspective is contains a range of predictions from positive to negative.
My Lifelong Interest in CAI
x
Conclusion
Efficiency is the essence of all successful business. Those businesses that will not automate everything within their power to automate will become too expensive to hire. That’s not to say that a human touch must and will disappear from business. That human touch is a value to customers. It is part of the numerator. If you squander your human touch by missing the right opportunities for your human employees to be human for your customers, or waste human touch by applying it when it’s not needed, you have just reduced your efficiency. Let your human employees be seen and heart. This is one reason I view conversational AI as a back-end function as much as a front-end one.
What Others Say about the Future of Business
Tao Yu, 2021 PhD Dissertation
All consumer-facing software will one day have a dialogue interface, and this is the next vital leap in the evolution of search engines.
Big Tech Says Big Things about the Future of CAI
This is why Microsoft has gone as far as to announce that the operating system of the future isn’t Windows, but “conversation as a platform.” Meanwhile, Facebook is pushing chatbots so hard that CEO Mark Zuckerberg says they’re the key for businesses that want to sell to Messenger’s 900 million monthly users.
http://www.talentzoo.com/digital-pivot/blog_news.php?articleID=22591
85% of Customer Interactions will be Handled without Human Agents by 2021 — Chatbots Life, 2019
Chatbots will change the world, that is for sure. No matter whether or not we prefer talking to agents, in the future we will be forced to rely on chatbot support. Chatbot statistics in this Gartner report predict that 85% of customer interactions will be managed without humans by 2021. This is good news. Consumers will get answers instantly, companies will reduce costs, and agents can spend their time solving more important issues.
https://www.smallbizgenius.net/by-the-numbers/chatbot-statistics/
Virtual Assistants will Get Deeper into Businesses — Kore.ai, 2019
With time, AI-powered virtual assistants or chatbots are expected to evolve into tools for enterprise business transformation beyond just being conversational interfaces. Enterprises using virtual agents for automating and addressing customer queries, will extend them to take action after customer requests. Enterprises can achieve the real value when conversational AI is used for business transformation by connecting conversations to business systems.” — Interesting points made by Mr. Sairam Vedam, Chief Marketing Officer, Kore.ai in an interview with Wire19. …
In near future, virtual assistants will be the first line of interaction with customers in the coming decade, just like the way the 1990s were dominated by the telephone call center and the 2000s saw the shift to websites and email.
Virtual assistants will form a synergy with other digital solutions such as IoT, analytics and image recognition, and will become an interface that compiles the information from these devices to offer meaningful insights to users. When clubbed with advanced analytics capabilities, VCAs can process information much faster than human agents and will predict consumer needs and behavior.
https://wire19.com/interview-with-sairam-vedam-cmo-kore-ai/
Bots are already Paying for Themselves—Accenture Survey, 2018
Although challenges remain, the future for bots is, indeed, bright. The potential benefits are simply too difficult to ignore. Just ask Amtrak, whose use of conversational bots has generated $1 million in customer service email cost savings annually — a reported 8x return. Or communications firm Charter, where bots cut customer service costs by 44 percent in the first year and generated a 5x return in just six months. Or soup ingredient maker Knorr, which saw a 50 percent increase in soup-stock cube consumption within just three months of launching a bot that gives recipe support to mothers. Or even companies in the Accenture survey that have already implemented bots, 91 percent of which expect to see a 1x to 5x return on their investment within the first 12 months.
Lego + Facebook, 2017
Lego Messenger Bot: Increasing sales conversions with a bot for Messenger
Goal: Increasing online sales. LEGO sought to boost sales by offering consumers gift recommendations that would guide them through the company’s vast catalogue.
3.4X higher return on ad spend for click-to-Messenger ads compared to ads that linked to the LEGO website.
71% lower cost per purchase when clicking through to the Messenger experience compared to ads optimized for clicks.
1.9X higher value for website purchases made from click-to-Messenger ads.
Conversational interfaces will be an important touch point in the near future. Being able to communicate and sell our products through Messenger has proven to be an innovative project and a commercial success. We are definitely happy with the initial results, and beyond excited to partner with Facebook on ideating further on the bot for Messenger for 2018. Lars Silberbauer, Senior Global Director, Social & Video, LEGO Group
https://www.facebook.com/business/success/2-lego
Jobs Going Away
When it comes to automation taking away jobs, a new report by Bank of America Merrill Lynch is predicting what amounts to near-total apocalypse: 800 million human jobs deleted by 2035. “Increasingly office and service sector tasks are being automated due to improvements in computing and software,” added the report… The price of robotics may decline 22 percent over the next six years, bringing pervasive automation within the reach of more firms.
https://insights.dice.com/2019/11/13/automation-apocalypse-tech-jobs-avoid/
Big Lists of Chatbot Stats
https://www.smallbizgenius.net/by-the-numbers/chatbot-statistics/
