Red Hat Inc. today announced the general availability of Ansible Lightspeed with IBM Corp.’s watsonx Code Assistant, a generative artificial intelligence service for information technology automation.

Announced in May, the service accepts prompts entered by a user and interacts with watsonx’s proprietary foundation models to produce code recommendations based on Ansible best practices. Purpose-built for Ansible, the service delivered a 25% to 40% productivity boost among automation developers during its pre-release phase, said Matthew Jones, chief architect of Ansible Automation at Red Hat.

The generative AI capabilities should be transparent to users, Jones said. When they create and edit Ansible Playbooks and rules, they can input a text prompt and receive an output that’s translated into YAML (which stands for YAML Ain’t Markup Language) code. Lightspeed with Watsonx also has the collateral benefit of supporting best practices in automation development and maintenance, Jones said.

“The training database is everything the community has submitted to our Galaxy product,” he said, referring to Red Hat’s repository of shared Ansible code. “That includes automation written by Ansible users, Red Hat and Red Hat partners. It’s tens of gigabytes of data.”

The result is that the watsonx Code Assistant reinforces current and proven coding practices. “The way people write Ansible has changed over the years,” Jones said. “We’ve developed best practices and style guidelines, and we want to be sure people are adhering to them.”

Transparent extension

Red Hat acquired Ansible in 2015 and has been steadily adding new features, including a generative AI front-end and streaming support last May. Ansible is an agentless open-source automation tool that simplifies complex tasks like configuration management, application deployment and task automation. It’s frequently used in infrastructure-as-code scenarios.

Red Hat said the watsonx copilot is a natural extension of existing Ansible workflows and content tools and that it delivers code recommendations that are accurate, consistent and specific to business needs. It’s natively integrated with the Ansible VSCode extension, meaning that developers and operators don’t need to log into or access a separate tool or service to access the Code Assistant capabilities.

Generative AI recommendations are also self-documenting, providing references to between three and five training sources for each suggested automation. Upstream content contributors can also choose whether or not their work contributes to model fine-tuning.

Red Hat has added a number of enterprise-oriented features since the initial announcement. “We have evolved the model a good bit, refined the service and added capabilities for cleanup and speed,” Jones said. Features targeted at commercial customers include enhanced data pipeline filters to clean up data, prevent leaks and preserve the context of the language model.

Ansible Lightspeed with watsonx Code Assistant is now generally available with an Ansible Automation Platform subscription. The service will carry an additional but as-yet-unspecified fee. Fine-tuning capabilities to train custom models specific to organizations will be available later this year.

Image: Pixabay