IBM Cloud Pak for AIOps is a robust and flexible IT Operations platform, connecting to 100s of different data sources, bringing together a holistic and unified view of a client’s operations and applying many sophisticated AI techniques to simplify an operators job.
To ease onboarding for new operators and to help our skilled operators delve even deeper, IBM Cloud Pak for AIOps is introducing a tech preview of the new Configuration Assistant in the 4.11.0 release. This is the first of many exciting agentic use cases to assist IT operations teams.
The Configuration Assistant has been trained using product documentation, support documents and best practices to answer an operator’s questions and to guide them to the best implementation. It uses a mix of Llama and Granite models for its implementation.
One example guides an operation wishing to integrate a new data source, on the best type of integration to use. Then reminds them to create some policies for it and finally assists on how to test this integration using some curl commands.Additional topics include how to use different functions within the product, such as runbooks or incidents, as well as how to configure policies, outbound integrations and more.
Another new exciting capability as a tech preview in the latest version is Incident Summarization with watsonx.ai. IBM Cloud Pak for AIOps uses the streams of events, deduplicates these into alerts then enriches these alerts with topology context, business impacts, anomaly detection and more. It then applies a sophisticated set of algorithms to prioritize and correlate these related alerts into an incident. This incident in turn is enriched with impacted applications, probable cause alerts, similar tickets, automations and other relevant data. This single view provides a real time savings to the operator and improves the mean-time to diagnosis.
Building upon this, IBM Cloud Pak for AIOps now includes the ability to generate an incident summary. This textual description uses the underlying incident data to quickly summarize the state of the incident so team members can come up to speed quickly as well as allow a simple summary to be shared with others. This summary can be customized if desired, and its history is included in the activity log to always have a summary of the state of the incident at a given time.
The incident summary uses a client’s own watsonx.ai instance so that the incident data stays under their control.
In closing, IBM Cloud Pak for AIOps is happy to introduce the first two capabilities using watsonx Granite models to continue to improve on the mean time to diagnose and to assist in answering any usage and administrative questions.
A special thanks to Jean Schunck, Pujitha Kara, Haibin Liu, Peter Gramenides and Franco Forti