Chatbot and automation: saving time and money (a public sector case study)
About the Client
Our client is a public administration entity. In response to the ongoing digitalization of document management and the execution of administrative processes, the client utilizes an electronic document management system (EZD). The goal of implementing this solution was to enhance the efficiency of system users’ workflows in a rapidly evolving and dynamic environment.
The challenge
- High volume of user inquiries directed to the technical department: Frequent system updates, new functionalities, and a rapidly growing number of EZD system operators impacted workflow efficiency and required frequent support from technical and subject-matter experts.
- Time-consuming processes: Employees had to review and analyze large volumes of messages, ever-changing documentation, and procedures, often seeking assistance from subject-matter experts. This slowed their work efficiency and consumed additional resources.
- Knowledge updates: Manually updating data sets and informing employees about changes in knowledge bases was time-intensive and required regular actions.
- Limited availability: Due to the limited number of personnel supporting operators and their restricted availability (other responsibilities, absences, etc.), some inquiries were handled with delays, while employees often attempted to resolve issues independently, leading to errors in system operations.
Key Project Objectives
The implementation of the new solution aimed to enhance work efficiency, focusing on the following goals:
Automating support inquiries: Enable the automated resolution of EZD system operators’ questions, reducing the workload on support staff and lowering operational costs.
High availability: Provide users with real-time access to necessary information and assistance.
Optimizing work time: Allow employees to quickly access the information they need, saving time spent reviewing documents and constantly evolving user manuals and system documentation.
Improving work quality: Enhance the efficiency of operators in document processing and case handling by delivering faster responses to their support requests and precise information. Significantly reduce errors that are difficult to reverse in the system by eliminating operators’ independent troubleshooting attempts when working with the EZD system.
Streamlining the EZD system onboarding process: Address the increased support needs of new groups of operators starting to use the system, especially for basic operations, and the repetitive nature of similar assistance requests. This was resolved by automating the delivery of immediate and accurate information to new users.
Pre-implementation situation
Before implementing the new solution, the client relied on methods that limited the speed and efficiency of handling technical inquiries, such as:
Traditional paper and electronic resources: Documents, procedures, informational materials, manuals, and system documentation were stored in both paper and electronic formats (e.g., in Intranet resources), making it challenging to quickly locate the required information. Additionally, these resources were subject to regular updates and changes.
Manual inquiry management: All inquiries, including frequently repeated requests and questions from operators, required repetitive, hands-on support from staff.
Manual knowledge base updates: Data sets were regularly updated manually, requiring employee involvement to implement changes and inform teams about new information.
Our solution
In response to the client’s challenges, we proposed a solution based on cutting-edge artificial intelligence technologies and AWS cloud services, enabling full automation of the support process and significantly improving access to information. Our implementation includes:
- Chatbot integrated with the knowledge base: We developed an intelligent chatbot capable of searching designated sources of information, such as documents, product websites, online publications, tutorial videos, and other multimedia materials. This allows users to receive accurate, real-time answers to their questions, ensuring up-to-date and relevant responses.
- AWS Integration: Leveraging AWS services like S3, OpenSearch Serverless, and Bedrock, we built a secure and efficient infrastructure for data storage and AI model training based on a closed knowledge base.
- 24/7 Availability: The chatbot operates around the clock, providing uninterrupted, instant access to needed information—even outside typical office hours or during times when EZD system support personnel are unavailable.
- Automated knowledge base updates: The system facilitates automatic updates to the knowledge base as new data becomes available, ensuring users always access the latest information without requiring manual input from employees. Changes in documentation can also be summarized and delivered to EZD operators through the preferred communication channel.
- Integrated communication channels: The chatbot can be seamlessly integrated with applications like Microsoft Teams, Slack, Messenger, or embedded as a widget in a specific application or website. This allows it to work fluidly alongside other systems used by the client, enhancing operational support.
- Versatility: The chatbot assistant can support various processes and systems within the organization—not just EZD—depending on the type of data supplied to its knowledge base, making it adaptable to diverse organizational needs.
Benefits for the Client
By adopting a modern solution powered by AWS cloud services and artificial intelligence, our client realized significant benefits, greatly improving the efficiency of document management, case handling, and reducing operational costs. The key advantages of the implemented system include:
Increased efficiency: Automation of the support process has significantly reduced the workload on employees. They no longer need to manually search through evolving documentation or wait for helpdesk assistance, which previously required the involvement of highly qualified personnel. This allows staff to focus on more critical and valuable tasks.
Reduced operational costs: Automated responses provided by the chatbot have minimized the need for additional staff, leading to lower costs for operating the EZD system.
Quick access to the latest information: Automatic updates to the knowledge base ensure that users always have access to the most current data. This reduces the risk of time-consuming errors and enables users to quickly learn how to use new system functionalities.
Improved user support and onboarding process: The integrated system provides instant responses and directs users to the correct information sources, resulting in better user experience and eliminating bottlenecks in helpdesk operations. These bottlenecks often arose during onboarding waves for new groups of EZD system users.
Enhanced scalability: The solution is easily scalable and can handle an increasing number of inquiries, ensuring long-term efficiency and adaptability to the growing number of operators being onboarded to the EZD system.
Tools and services used
To build an efficient, secure, and scalable solution on AWS cloud, we leveraged a variety of services that facilitated effective automation of inquiry handling and seamless integration with other platforms. These included:
- AWS S3 (Simple Storage Service):
Employed for secure data storage, including documents and other resources leveraged by the chatbot to deliver accurate responses to user inquiries.
- AWS IAM (Identity and Access Management):
As a means of ensuring full access control to system resources, enabling secure management of user and administrator permissions.
- AWS VPC (Virtual Private Cloud):
Communication within the system takes place over a private VPC network, enhancing security and minimizing the risk of unauthorized access to data.
- AWS OpenSearch Serverless:
Used as a vector database for artificial intelligence models in a serverless architecture, enabling fast and efficient real-time query processing.
- AWS API Gateway:
Used to integrate the chatbot with external services such as Microsoft Teams and Slack, enabling seamless transfer of queries and responses across different platforms.
- AWS Lex:
A service that enables the creation of a chat interface, allowing users to directly interact with the AI model, which processes queries and provides responses.
- AWS Lambda:
Used to execute code and individual components of the solution upon receiving a query from the user interface, ensuring flexibility and automation of processes.
- AWS Bedrock:
Utilized for training the artificial intelligence model using an internal, closed knowledge base, enabling efficient querying by the chatbot.
