The Ultimate Enterprise Chatbot Platform
When creating a bot, and when creating any software for that matter, there are different considerations for enterprises as opposed to small businesses.
While both the enterprise and the smaller business will strive for creating the best possible bot with the resources and within the time available, the enterprise has many more issues to consider.
The first is security.While a smaller business needs to be concerned about security and must take appropriate precautions and have a security policy in place, an enterprise has much more at stake. The costs of a security breach can be much higher in both financial and brand terms than for a smaller business. The data belonging to large enterprise is obviously much more of a target for hackers than a small company.
The second is brand. Any solution deployed by an enterprise that performs poorly has the ability to impact the brand. This is particularly true because an enterprise is a higher profile company with minimum performance standards in the mind of the customer. Any missteps can result in unwanted publicity. This is even more of a risk with artificial intelligence.
The third is scale. An enterprise will have a larger customer base and therefore needs to implement solutions that can scale performance comfortably to the required level of engagements. The number of connected mobile devices and the ability to draft off social media platforms, makes performance an important issue.
The fourth is complexity of teams and systems.Smaller businesses often have a single person or single team implementing a new technology. An enterprise on the other hand often has multiple stakeholders involved in the release of a new product and multiple teams actually working on the implementation. In addition the systems are more complex, especially when they involve machine learning, and therefore the negative impact of bad software design choices is likely to be higher for an enterprise than for a smaller business. Interestingly the complexity of enterprise systems actually creates an opportunity to build a technical support chatbot to help individuals or teams achieve tasks.
The fifth is control. Enterprises often have the internal resources to be able to optimize any aspect of a system or data. Whether a small enterprise may have the goal of creating a great chatbot for marketing, business, or ecommerce, the enterprise will likely want to have full control of its data and source code so that over time it is able to use its internal development capabilities to increase synergies between applications and services.
The sixth is competition. The enterprise is likely to face tougher well funded and global competition. The enterprise bot will not only be judged on its own merits but will be judged against the high profile competition.
Chatbots developed for enterprise need to address the above considerations. Let's examine these considerations one by one.
An enterprise bot needs to be created with security in mind.
One of the first steps an enterprise can take in this regard is to host the bot on-premises. Hosting the bot on premises means that the enterprise will have full control of the servers that it runs on and of the development process. It can therefore implement its full security policy.
Besides ensuring that the conversations with the customer are suitability encrypted, the enterprise will also need to implement features like role based security and multi-user management so that access to the bot functionality can be controlled to a high level of granularity.
The first consideration for an enterprise is that anything that they release does not detract from their brand.
They however will need to ensure that the experience of using the chatbot actually enhances the brand. To enhance the brand, the enterprise chatbot needs to remove as much friction as possible in allowing the customer to complete the job at hand and at the same time be a pleasure to use.
The chatbot must also be aesthetically pleasing and consistent with the brand aesthetics. This can be an issue at the moment as chat platforms have limited features to control branding.
We believe however that this will change as chatbots become more graphical by adopting more graphical user interfaces and widgets.
The enterprise will also need sophisticated and customizable analytics to ensure to be able to monitor and refine the brand experience. Analytics can help identify problems and opportunities. Using market segmentation, A/B testing and other analytics they can not only capture valuable information about their customers, but they can also personalize the chatbot experience for the customer in question, potentially further enhancing the brand experience.
The enterprise must be sure that they have the ability to scale the performance of their systems to ensure that the chatbot can perform at a high volume of interactions, particularly that the chatbot can handle a high volume of simultaneous interactions.
To ensure this can happen, not only must the underlying framework support the architecture required to process multiple interactions in parallel, but the framework itself must be capable of the performance levels required by enterprises.
The framework should have the necessary stress testing and analytics tools built in to ensure that any required performance levels can be planned and tested in advance.
Complexity of Teams and Systems
It must be possible for the enterprise to manage the various teams involved in the implementation in an efficient and secure way. For example, the content needs to be managed by the content team (pre and post deployment) and the flow needs to be created and managed by the developers.
This type of role separation needs to be implemented in the framework in three ways:
- Firstly at the development level is should be possible to separate the content and flow so that they can be created and managed independently of each other. Botpress invented Universal Message Markdown for this purpose.
- Secondly, access to the system and its interfaces pre and post deployment needs to be managed using multi-user management and role based security. Moreover they would likely want to be able to monitor and manage all their bots from one dashboard.
- Thirdly, by having an elegant and extensible architecture the framework makes it easier for team members to easily understand where components and functionality fit in, even if they weren’t involved in the original implementation. In addition, the systems complexity is reduced by having a framework that enforces and promotes good design.
To have control over the code and data the enterprise must have access to the code and data. While it is obvious to point this out, many bot solutions do not give enterprises control of the source code and the data which limits their flexibility. With full control of the data and source code the enterprise can fully integrate and customize any bot and also share common functionality (such as payment capabilities) between bots.
An enterprise needs the bot to be positivity differentiated against the competition. To achieve this the enterprise has to use the framework that gives them the flexibility and efficiency to create the best customer experience with the resources and in the time scale required.
All of the above considerations potentially differentiate an enterprise bot from a bot created to provide services for a smaller business.
While all the above considerations are important and must be addressed, of course the main priority is to develop a bot that succeeds in offering the customer a much better way of accomplishing the job at hand. This is not the exclusive goal of the enterprise, as smaller businesses are just as concerned about creating great customer experiences, however it is a point that must be emphasized.
There can sometimes be the tendency when a new technology emerges to embrace the tech to the extent that the customer experience is unintentionally deprioritized. While this danger exists for all companies, it might be harder to catch within enterprises because it’s harder for managers in large organization to monitor every detail of what is going on.
One particularly dangerous space for enterprises in the “chatbots for enterprise” space is the use of AI. While amazing advances in Natural Language Processing (NLP) have meant that the speaker's intentions in natural language can be identified with a high degree of confidence, there is an erroneous perception that AI more generally is able to carry out deeper conversations with customers about their products. While these types of systems may be viable in future, right now trying to get a chatbot to understand context, ambiguity and have a meaningful memory is a quixotical quest that can suck up endless resources and time.
Enterprises need to consider many issues when building a chatbot experience for their customers. And that is in addition to making sure that the chatbot solves a real problem for the end customer and that the experience is truly great. Even if the enterprise starts by building simple prototypes that it rolls out to a limited customer base, there are still important issues to consider including ensuring that the prototype is built on a framework that will make it easy to proceed towards production once the relevant information has been gathered.
Enterprises are beginning to embrace chatbots and we believe that enterprise bots will soon become mainstream channel for customer interaction.