2024 Gartner® report: Tips to bolster your disaster recovery program
No items found.

Cookie consent

By clicking “Accept”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.

November 15, 2023

Build runbooks in minutes, not hours with AWS Bedrock and generative AI

We all know that generative artificial intelligence (GenAI) is rapidly evolving and accelerating content creation. However, for enterprise IT operations, using GenAI without proper training on structured and unstructured data can get the process flow entirely wrong.

In our innovation labs at Cutover, my team has been experimenting with various ways to leverage GenAI tools including those provided by AWS such as Sagemaker, Bedrock and their large language model (LLM), Titan, to create Cutover runbooks from unstructured data sources. Using these services we have been able to quickly iterate to prove we could auto-generate runbooks, as well as suggestions for the best next steps to take when supporting IT operations ranging from cloud migrations to IT/cyber disaster recoveries.

As an example, we built a prototype to pull from a Confluence setup of 200 AWS best practices. In this prototype we were able to demonstrate how we can offer customers AI generated runbooks from single line prompts. In addition, we are fine tuning models to include best practices and suggested courses of action, such as when a workflow is “stuck”, we can use GenAI to suggest and dynamically create a new work stream to unblock it.

Areas where we have found quick success using AWS Bedrock and Titan include:

  • Creating runbooks and tasks at an unparalleled speed. For example, we used AI models to create guidance on how to migrate an on-premises database to AWS Aurora in a matter of minutes. This process flow would normally take hours, if not days, to construct, especially if you were looking at the complete guidance set of thousands of blog posts, quick starts and development kits available from your cloud provider.
  • Developing suggested approaches via sentiment analysis by reducing complex choices that might number in the thousands to a single weighted and risk-analyzed choice.
  • Training our models to generate real-time suggestions in runbooks to make informed and conditional choices for the remaining tasks to accelerate to the next key milestone.  

I am excited to see how Cutover is leading the way on how to use GenAI to make technology operations solutions more efficient and productive and get the best from existing knowledge bases. If you are interested in learning more about what we are doing with GenAI, specifically AWS Bedrock and Titan, feel free to reach out to me at info@cutover.com. If you’re attending AWS Re:Invent on November 27th-December 1st, let’s chat there.

Kieran Gutteridge
Collaborative automation
Latest blog posts