ChatBot – An effective tool for Enterprise Consumerization and operating lean in the times of CoVID
Chatbots are defined as programs designed to hold conversations with humans. Chatbots are now being implemented in numerous spheres to automate and streamline processes, drive productivity improvements, and improve user experience.
This PoV describes our experience and learnings from the implementation of a Chatbot for streamlining internal workflows.
The client, a leading digital transformation organization, has grown rapidly in India. This rapid growth meant that business enabling functions like staffing, finance, facilities, and human resources had to scale in terms of the services provided to employees. These functions manage incoming, present, and outgoing employees while supporting them on multiple issues.
The business-enabling functions enhanced their internal processes and added specialists to address the problem of scale. They expanded and communicated information proactively through channels like onboarding workshops, helpdesks, information broadcasts, etc.
However, the information dissemination was limited by employee attendance and varying interests. In addition, the constant flux of employees implied that there are always new employees asking the same set of questions. About 60% of the questions were mundane and repetitive.
The functions desired a Chatbot solution to automate answering frequently asked questions. This solution was expected to save many people hours of daily effort for these areas and help employees with instant answers, maintaining the necessary human touch.
The ChatBot would answer questions including:
- How do I access the Holiday calendar?
- How do I apply for leave?
- What should I do if the claim & hard copies are submitted but payment not received?
- Where can I find the Travel Policy?
- When can I make changes to my salary structure?
- How do I submit expense reports?
The solution had to work for 1000+ employees spread across 10+ departments and had to integrate with existing remote working tools from GSuite to provide a seamless experience to employees.
The need for such a solution intensified further due to the COVID-19 outbreak resulting in the need to digitalize internal operations as soon as possible. With almost 100% of employees working remotely, it was imperative to implement a solution to reduce overheads and dependencies on support functions and improve the employee experience.
The solution is built using Google DialogFlow as the company has existing Google Cloud Platform subscriptions, limiting the dollar investments, and it integrates seamlessly with remote working tools in the company.
DialogFlow is a google service for creating conversational interfaces for websites, mobile applications, and popular messaging platforms. It can be used to build interfaces that enable natural and rich interactions between users and businesses.
Basic building blocks of DialogFlow are agents, intents, entities, and contexts. An agent is a virtual one that handles conversations with end-users. It has the ability to understand natural language and translate it into intent and parameters of intent. One can define an action for each intent and use the parameters to respond to the intent. Context is similar to natural language context and the responses use the context for an appropriate reply. DialogFlow provides out of the box integrations with many popular conversational platforms such as HangOut, Slack, Facebook, etc.
A DialogFlow Agent was created along with its internal constructs like intents, entities, and constructs. The Agent was trained with knowledge base documents from various functions. HangOut chat integrations were enabled to make the Bot available on internal messaging platforms.
Furthermore, an Agile approach was employed to build the solution with initial sprints focusing on technical viability and stakeholder buy-in. Once the stakeholder buy-in was established, the Chatbot functionality was built incrementally, adding new domains and training data in each sprint. The performance of the model was analyzed and fine-tuned to improve the accuracy of the answers.
Key lessons learned from the implementation are:
- Most likely enterprises have existing Chatbot solutions. Analysis has to be done to align new solutions with existing assets and establish the value addition.
- There is a high probability to find existing and unused “dead” Bots. Implementation of Chatbots is often considered a one-time activity without any considerations to the requirement of monitoring and tuning of the ML models powering them. Another factor for the “death” is the lack of integration of the solution into processes and communication to users.
- It is important to explain the possibilities and technology limitations to the Lines of Business (LOB)/functions/SMEs to evaluate possible solutions. Knowledge of these stakeholders is often influenced by a sales pitch or media rather than available technical capabilities.
- Mechanisms have to be incorporated to continuously analyze, train, monitor and tune the performance of the Chatbot. Some of the issues that may have to be addressed:
- misunderstanding of requests
- tuning of parameters of classification models
- unanswered questions
- inaccurate answers
- similar questions in multiple knowledge bases
- soliciting feedback from user groups
- Service providers / Platforms offer multiple features out of the box that improves the end-user experience. Examples are Small Talk, automatic spell check feature, Analytical reports, Model validations.
- Efforts to gather and collate information must not be underestimated, particularly when the BOT answers questions from multiple sources. Coordination mechanisms have to be established to organize and maintain data.
- Chatbot solutions must be promoted aggressively and incorporated into internal processes and communications for it to succeed.
About the Author:
Satish Peyyety has over 19 years of experience leading large scale application development projects in emerging technologies. He is an avid learner of technology, cleared 15+ certifications in the areas of Cloud, Java and IT Architecture. Presently he manages the Cloud & DevOps Studio in Globant India and leads exploration and development of accelerators and tools.
A digitally native company where innovation, design, and engineering meet scale, Globant is a disruptor in the professional services arena. We leverage the latest technologies and methodologies in the digital and cognitive space to help organizations transform in every aspect.
A pure-play in the digital and cognitive fields, Globant has a global presence with delivery centers in North America, Latin America, Europe, and Asia through which we offer a deep knowledge in the latest trends and technologies. With over 12,000 professionals present in 17 countries working for companies like Google, EA, and BBVA among others, we are featured as a business case study at Harvard, MIT, and Stanford apart from being a member of the Cybersecurity Tech Accord (2019).
For more information, visit www.globant.com