Skip to main content

Release Agent

Last updated on

Release Agent is an in-app AI chatbot designed to streamline the use of Harness Feature Management and Experimentation (FME). It offers multilingual support, rapid responses, and knowledge-based assistance by using Harness public documentation and blogs. Release Agent makes it easy for all developers to get the help they need without leaving the Harness interface.

What will you learn in this topic?​

By the end of this topic, you will be able to:

Before you begin​

Before you use Release Agent, ensure you have the following:

  • Harness FME access: Access to the Feature Management and Experimentation module.
  • Admin permissions: Admin role to enable or disable Release Agent for your team (required only for configuration).
  • Active feature flags or experiments: At least one feature flag or experiment configured to use AI Summarize features (optional for Ask Release Agent).

Ask Release Agent​

Ask Release Agent provides conversational assistance for FME tasks and workflows. Navigate to the FME navigation menu and click Ask Release Agent to open the chat modal.

Example questions​

Release Agent can answer questions across setup, configuration, best practices, and troubleshooting:

  • How do I set up a JavaScript SDK in my codebase?
  • How do I create and manage feature flags in FME?
  • How can I integrate FME with my application using the SDK?
  • What are the best practices for naming feature flags and events?
  • How do I set up an A/B test in Harness FME?
  • Can you explain how FME's targeting rules work?
  • How can I use segments to target specific groups of users?
  • What data types can I track using FME events for measurement?
  • How do I interpret the metrics on my experiments' results?
  • How do I export data from FME for further analysis?
  • What are the implications of feature flag rollout strategies, such as canary releases?
  • How do I troubleshoot issues with the FME SDK?
  • Can I use FME across multiple environments, and how?
  • How does Harness FME handle data privacy and security?
  • What are best practices for cleaning up old flags and technical debt?

AI Summarize​

AI Summarize analyzes and summarizes metric results on feature flags and experiments. This feature helps you understand performance data and experiment outcomes without manual analysis.

Feature flag metrics​

To use AI Summarize on a feature flag metric, drill into a metric tile on a feature flag's Metrics impact dashboard and click Summarize.

Experiment results and metrics​

AI Summarize works on both individual experiment metrics and overall experiment results.

To summarize an experiment metric, drill into a metric on an experiment's Metrics dashboard and click Summarize.

To summarize an experiment's overall results, select an experiment on the Experiments page and click Summarize.

Go to Viewing experiment results for more information on AI Summarize.

Enable Release Agent for your team​

Admins can enable or disable Release Agent and control experimentation data processing through FME settings.

  1. Navigate to FME Settings > AI settings.
  2. Select the Enable Release Agent toggle to turn the chatbot on or off for your team.
  3. Select the Enable processing of experimentation data toggle to enable or disable AI Summarize features. When you enable processing of experimentation data, the data is protected by the privacy policies described in the Privacy section.
  4. Click Save.

Your data usage​

Release Agent processes data with accuracy and privacy standards to ensure secure and reliable AI assistance.

Accuracy​

While Release Agent strives to provide accurate information, it cannot guarantee the accuracy, completeness, or timeliness of all information. Large Language Models (LLMs) can struggle with precision and context of numerical information.

You can reach out to support@split.io if you have questions or require additional help.

Privacy​

Your data is protected by the Harness privacy policy, which applies when you use the product. Release Agent uses OpenAI as a data subprocessor for the data summarization feature, so your data is additionally protected by the OpenAI Enterprise privacy policy.

Additional ways that your privacy is protected:

  • No training on your data: Neither Harness nor OpenAI uses your data for model training.
  • Minimal data sharing: Harness sends OpenAI only the minimal necessary information required to create data summarization.

Feedback and support​

Submit support inquiries or report issues with Release Agent via support@split.io.

Provide detailed descriptions and reproduction steps for issues to facilitate expedited troubleshooting. Your feedback is instrumental for the iterative enhancement of Release Agent.

Next steps​