Learning Library

← Back to Library

Accelerating Ansible with Watson X

Key Points

  • IBM Watson X Code Assistant for Red Hat Ansible LightSpeed uses generative AI to turn natural‑language prompts into Ansible playbooks, allowing users to install and configure services like Apache with a single command.
  • Users can combine multiple tasks in one prompt by prefacing the prompt with a hash and separating instructions with ampersands, then accept or edit AI‑generated recommendations via a tab key.
  • The platform exposes the training data sources (author, license, etc.) for transparency and lets organizations fine‑tune the underlying large language model with their own private Ansible data to produce more relevant, customized code suggestions.
  • An Ansible Content Parser converts existing playbooks into a single JSON‑L file, which can be uploaded to the intuitive Watson X Tuning Studio where non‑experts can run tuning experiments, view metrics, and monitor training loss graphs as the model improves.
  • After tuning, the customized model is deployed by copying its ID into the admin portal, making the enhanced suggestions available to authorized users through Red Hat’s Ansible VS Code extension.

Full Transcript

# Accelerating Ansible with Watson X **Source:** [https://www.youtube.com/watch?v=vG54TKeEwP4](https://www.youtube.com/watch?v=vG54TKeEwP4) **Duration:** 00:03:48 ## Summary - IBM Watson X Code Assistant for Red Hat Ansible LightSpeed uses generative AI to turn natural‑language prompts into Ansible playbooks, allowing users to install and configure services like Apache with a single command. - Users can combine multiple tasks in one prompt by prefacing the prompt with a hash and separating instructions with ampersands, then accept or edit AI‑generated recommendations via a tab key. - The platform exposes the training data sources (author, license, etc.) for transparency and lets organizations fine‑tune the underlying large language model with their own private Ansible data to produce more relevant, customized code suggestions. - An Ansible Content Parser converts existing playbooks into a single JSON‑L file, which can be uploaded to the intuitive Watson X Tuning Studio where non‑experts can run tuning experiments, view metrics, and monitor training loss graphs as the model improves. - After tuning, the customized model is deployed by copying its ID into the admin portal, making the enhanced suggestions available to authorized users through Red Hat’s Ansible VS Code extension. ## Sections - [00:00:00](https://www.youtube.com/watch?v=vG54TKeEwP4&t=0s) **Watson X Ansible Code Assistant** - IBM's Watson X Code Assistant for Red Hat Ansible LightSpeed uses generative AI to turn natural‑language prompts into playbooks (e.g., installing Apache), provides tab‑accepted content recommendations with visible provenance, supports multi‑task prompts, and can be customized with an organization’s private Ansible data for tailored outputs. ## Full Transcript
0:00IBM Watson X code assistant for Red Hat 0:03ansible light speed leverages generative 0:05AI to accelerate the creation of anible 0:07playbooks and helps organizations 0:10implement it automation let's see how it 0:12works by asking Watson X code assistant 0:14to install and configure an Apache web 0:17server to start simply input your 0:19commands in natural language and press 0:21enter you'll quickly be served in AI 0:23generated content recommendation if you 0:25want to combine multiple tasks within a 0:27single prompt Begin The Prompt with a 0:29hash 0:30and add ampersands between the different 0:32sets of instructions once again Watson X 0:35code assistant provides an AI generated 0:38content recommendation press tab to 0:40accept the 0:41recommendation or you can modify as 0:43needed to help Foster trust and 0:45transparency the training data that may 0:47have informed the content recommendation 0:49like author and license is easily 0:51accessible since every organization is 0:53different Waton X code assistant 0:55empowers you to customize its underlying 0:57large language model with your existing 0:59an data that helps deliver content 1:01recommendations that are crafted for 1:03your specific preferences for example 1:05say you're creating an open shift 1:07cluster when you enter your prompt 1:09Watson X code assistant suggests 1:11standard content from anible built-in 1:13module however when you customize the 1:15model with your organization's private 1:17anible data set the model generates 1:19custom content more aligned to your 1:21needs the anible content parser tool 1:23empowers developers to transform 1:25existing playbooks into a single Json L 1:28file filled with tr training data with 1:30the data all in one place you can easily 1:32create and run a tuning experiment in 1:35the Watson X code assistant tuning 1:37Studio as you navigate the tuning Studio 1:39the intuitive user interface provides 1:41helpful guidance in other words you 1:43don't need to be a data scientist to use 1:45it simply enter an experiment name and 1:49brief description you can upload your 1:51Json L file or simply drag and drop it 1:55once uploaded you can view your data's 1:56metrics and compare them to the models 1:58Baseline for instance here you see your 2:01data samples modules and unique modules 2:04and how they compare to the anible data 2:06used to train the base model you can 2:08drill into the module count to see the 2:10overall distribution of your data by 2:12module here the new modules are now part 2:14of the training data to initialize model 2:17tuning click start tuning you'll see the 2:20tuning status tracked on screen after 2:22the tuning is complete a training loss 2:24graph reveals the accuracy of your 2:26models predictions as compared to the 2:28training data the graph updates as you 2:30run multiple tuning Cycles training loss 2:33tends to decrease over time okay now 2:35that the customization is complete you 2:37can deploy your tuned model and use it 2:39in Watson X code assistant for Red Hat 2:41anible light speed simply copy the model 2:44ID into the admin portal when activated 2:47your model is accessible to your 2:49authorized users in the anible vs code 2:51extension by Red Hat with the tuned 2:54model Watson X code assistant will 2:56recommend content using modules 2:58functions and other details specific to 3:00your private it environment now let's 3:02head back to VSS code and run the same 3:04task again to see how Watson X code 3:06assistant returns new and improved 3:08content recommendations this time the 3:10module IBM container cluster is being 3:13recommended instead of anbl built-in 3:15module by providing AI based content 3:17recommendations IBM Watson X code 3:19assistant for Red Hat anable light speed 3:21helps make content development easier 3:23and more efficient for hybrid Cloud 3:25developers to see how it can help begin 3:27a trial or reach out to an IBM 3:29representative and book a live 3:45demo