The Coming Revolution: Chatbots

We are planning a series of posts about the interesting advances in Chatbot technologies, and our specific experience with them over the past few months. At PalGeek Inc. we have come to the realization that we are on the cusp of major disruption due to the advances in this technology, and we will share our insights in this up coming series of posts on Linkedin.

What Are Chatbots?

Chatbots are solutions with both hardware and software components that utilize natural language processing, artificial intelligence and machine learning to interact with human beings in a natural way. The best of Chatbots are the ones that can pass off as a real human, and ones that can learn from experiences.

What Is The Big Deal?

We will discuss how Chatbots will disrupt technologies and companies in a major way in our next few posts on this topic. However, simply, since computers and information systems will be able to interact with humans in a more natural way, and learn from experience the way that enterprises engage and create value for customers will change. Not only is the opportunity there to reduce operating costs and boost customer engagement, but there is a real opportunity to create new revenue streams and business models. Companies with no clear plans on how to use this technology will quickly get disrupted by those who do.

Why Now?

While the concept of Chatbots has been available for over a decade, a number of factors are helping this technology become commercially relevant:

Cheap computing power

As the cloud is driving the cost of computing to levels approaching zero, it becomes more affordable for SMEs to consider this important technology as part of their digital presence. In fact what we are seeing is that SMEs are more likely to adopt such a new technology because it further reduces their costs, improves customer engagement and provides them with a decisive edge over much bigger competitors.

A.I. as a service

For those who want to dabble with Chatbots, a number of companies are offering A.I. technology as a service with integrations into Facebook messenger or websites.

Advances in Machine Learning

Machine learning is a discipline in computer science allowing computers to learn from experience. This is an important requirement for any commercial Chatbot, that would learn from past interactions and whose behavior can be adjusted as needed.

Advances in Natural Language Processing

Natural language processing has always been one of the most challenging aspects about dealing with computers. Even the way that we search for content today using Google search engines for example is based around short key phrases. These keyword phrases are not the natural way we would communicate with each other and can make it challenging to find the information we are looking for. A natural language processor is capable of understanding the structure of human language and attempt to determine intent and derive complex meaning.

Investment

As governments and investors are pouring more money into A.I. companies and academic research in the areas of machine learning, natural language processing and Chatbot technology, the cost of development is going down, and the availability of skillets in this area is improving.

Multi-language Availability

A major deployment challenge for chatbots has been the support for non English language Chatbots, especially languages that do not have the same structure as English (even if these languages shared the same character set). However, over the past few months we have seen companies develop chatbots that support different languages, including Arabic for example.

More Integrations

As more applications are available in the cloud, Chatbots today have an array of APIs they can integrate into: This includes applications such as Slack, IFTTT, WordPress, YouTube, Instagram, Twitter, Location Services, Google Sheets, Zapier, and payment gateways. With time, more interesting integrations will become available allowing for versatile services and engaging customer experience.

Availability of On-Premise Solutions

For financial services, government and telecom providers having an on-premise solution is a matter of going for or completely skipping on the technology. Due to the many of the above factors such as cheap computing power, and improvement in the software quality, on-premise solutions are becoming available allowing enhanced security and control.

Acquisition Costs

Because of the interest we are seeing from private and public entities in this new and disruptive technology, the cost of acquisition is significantly coming down. It can range from DIY using an A.I. as a service platform, to fully fledged software solution starting from around the $100K mark. A very cheap technology in comparison.

Challenges in Chatbot Technology

While advances in Chatbots, and falling costs are very encouraging, a number of commercial and technical challenges remain.

It is still a new technology

Machine learning and natural language processing are still relatively new disciplines in computer science. Finding talent with the right skill sets maybe expensive, or unattainable. Chatbots are live programs that learn with time and need regular maintenance and development. This is a great opportunity for A.I. as a service companies that can fill this knowledge and services gap for large enterprises.

The technology is rapidly evolving

While this is great and exciting from a technical perspective, it is a challenge from a commercial perspective. An enterprise may end up buying the lower end model only to be out gunned by those who purchase more advanced models just a few months later. To navigate this risk ensure that software upgrades / more features are negotiated into the commercial contract. It is worth it.

Lack of common technology platforms

While most of the commercially available Chatbots today started from open source academic projects, they have been developed into highly proprietary pieces of software. This is a challenge from a commercial perspective because if as an enterprise you end up investing in a dead end path it maybe very difficult to recover your costs. You would end up having to build new Chatbots from scratch which is not only a commercial loss, but also expensive from an operating perspective and a loss of acquired Chatbot knowledge that would be impossible to transfer to a new Chatbot seeamlessly.

Hopefully a truly open source, widely adopted machine learning, and natural language processing platform will be available soon.

If you are interested in try a Chatbot yourself, Hubspot has built an interesting Chatbot called Growthbot which is intergarated into Facebook Messenger. Just send a message and say "Hi" to @Growthbot. We are also building ower own Chatbot on Facebook Messenger. Dropby and say "Hi" Frank at @BestWebsiteStore on Facebook Messenger. Our little guy is still early in the development stage, so don't expect too much out of him yet.

Our next post on Chatbots: 5 Ways Chatbots Will Disrupt The World