The challenge
About a year ago, a major telecom network outage in Canada affected major institutions such as banks and retail stores during payment processing, along with consumer internet and mobile data.
As a result, the national Tier 1 telecom operator invested in building AI-based troubleshooting and observability solutions.
The client's senior network IT operations team had too much operational log data arriving daily and was unable to fully use these error logs in outage scenarios.
The proposed solution
We proposed a solution that used an open-source data pipeline, machine learning and GenAI capabilities to stream this data into an intelligent AI assistant.
We built the solution around human-centric design and modelled a day in the life of these senior engineers.
We tested several leading LLMs and engineered the prompts that delivered the highest quality responses.
Expanding the AI capabilities
We can understand syslog formats from leading routers, switches, firewalls and load balancers.
We soon realized that error logs were a rich source of information that could benefit from additional ML-based analysis such as anomaly detection and correlation.
From pilot to production
After a successful pilot, this solution is going to be rolled out into their production network.
We have continued to develop the solution on our own, adding advanced dashboards and early warning insights that go beyond traditional monitoring solutions.
Expanding across the application stack
We are now adding the entire application stack to our library of log formats, from servers to APIs.
We continue to focus and innovate so that our platform can scale to thousands of devices and millions of logs per day.
Secure deployment
The platform is available for secure on-premise data centers and in your private cloud.