ServisBOT provides a Conversational Artificial Intelligence (AI) Platform for businesses to build natural language solutions, virtual assistants and chatbots that transform customer and employee engagement across multiple digital channels and across many different journeys.
Chatbot and natural language solutions now enable people to search for knowledge, raise tickets, query transactions, renew memberships, book travel, schedule appointments, transfer money, and engage across multiple journeys. This is what is known as Conversational AI, and it is the key to advance automation and self-service for customers and employees.
Transforming to a Modern Application
John Rellis is a team leader at ServisBOT. Their teams are based in Boston, Dublin, and Waterford. As a chatbot platform, their back-end application handles a massive amount of events, as their customers are utilizing their AI to build and run chatbots. These chatbots operate across a broad set of industries and handle high traffic that is often spikey.
“As a chatbot platform, ServiceBOT’s back-end application handles a massive amount of events, as their customers are utilizing their AI to build and run chatbots.”
In order to handle the high, unpredictable scale, the ServisBOT team chose to use containers and serverless architectures on AWS. They use native AWS services combined with modern services such as ECS, Lambda, SNS, AppSync, API gateway, and DynamoDB.
Having a cloud-native mindset enables ServisBOT to develop new features at pace while handling the quickly growing scale of their platform as they serve more and more customers.
Wanted: Reveal Unknown Errors
As the complexity of ServisBot’s cloud-native architecture grew, it became increasingly difficult to guarantee the high level of uptime required for seamless user experience. Far too many times in testing, messages would go unanswered and the Bots just could not make their minds up.
The seemingly random degradation of performance led to a lot of time wasted in Cloudwatch tracing errors through Lambda, SNS and Dynamo DB. That’s when they started to look into monitoring tools.
The main requirements they sought from a monitoring tool were notifications about errors and traceability. They were also interested in cost monitoring and specifically, the cost of each transaction. That’s when they found Epsagon.
“The main requirements ServisBOT sought from a monitoring tool were notifications about errors and traceability. They were also interested in cost monitoring and specifically, the cost of each transaction.”
Tracing Through the Entire Architecture
ServisBOT heard about Epsagon through AWS. They liked the fact that Epsagon’s product was already in general availability and they could start working with it right away.
Epsagon’s main technology was tracing, which the developers loved. It also required minimal setup. After getting through a basic configuration, they set notifications to Slack using Epsagon. Once set, the developers did not have to go into the AWS monitoring console anymore, which made development and troubleshooting faster and easier.
ServisBOT still uses its own mechanisms to correlate messages through the system from time to time, but Epsagon has reduced the burden.
“Its ‘cool’ to pull up the architecture view and show people. I even used the architecture view to show my wife what I was doing,” said John.
“Its ‘cool’ to pull up the architecture view and show people. I even used the architecture view to show my wife what I was doing.”
Up and Debugging in No Time
With an intuitive user interface and fast setup, ServisBOT was up and running on Epsagon in no time. The team’s primary use of the product is to get notified of errors in real-time and troubleshoot them faster than ever before. Previously, the team was losing valuable time and productivity using AWS tools. Another feature used by the team is Epsagon’s tracing functionality. With it’s ‘cool’ architecture and notifications on errors, Epsagon gave ServisBOT’s team confidence and reduced debugging time significantly.
“With it’s ‘cool’ architecture and notifications on errors, Epsagon gave ServisBOT’s team confidence and reduced debugging time significantly.”