Galileo AI

What is Galileo AI?

Galileo is an AI layer that sits on top of the LogRocket platform. It combines information about how users react to problems with traditional error reporting and analytics to develop a human-like understanding of user behavior and uncover high-impact, actionable insights in user behavior, including:

• Severe technical issues such as errors, failed network requests, and error states.
• Usability issues leading to user struggle such as rage clicks, dead clicks, and frustrating network requests.
• Issues causing users to drop out of funnels and key workflows.

How Does Galileo Work?

Galileo's models have been trained on billions of data points to predict whether identified issues and friction points are important, automating the analytics work that humans already do in LogRocket. Importance is based on vectors such as impact, frequency, and years of user feedback around what matters most.

Galileo learns via user feedback, so its recommendations are constantly improving. Activities such as triaging issues as "high impact", "low impact", or "ignored" help Galileo better understand what matters most and make more accurate and relevant recommendations in the future.

Galileo AI Features

Galileo scans your issues, assessing issue events for significant user impact that can be distinguished by user expressing frustration or confusion, or an explicit errors or error state disrupting and ending early a user journey.

Reduces noise, focuses on signal

Similar solutions create an incredible amount of noise and expect users to make sense of the noise, sifting through to look for signal. This is time- and resource-intensive for teams. Galileo directs users to the signal, filtering out the noise and leaving users with a short list of just the most impactful issues to save you time searching. It's not common for applications to have 5-15,000 issues, each with multiple issue events. Therefore, Galileo starts by reducing the a highlighted list of the 5-25 most impactful issues.

📘

Minimum issue event count to be severe

Issues need to occur a minimum of 3 times before LogRocket will consider an issue severe.

Issues Digest

Receive daily or weekly updates on the top recommended issues as identified by Galileo, delivered via Slack, Microsoft Teams, email, or webhook. Issues Digest will send the top 5 issues each week so that you can be efficient with your time and focus your efforts on the issues with the greatest impact.

Natural language issue titles (in Beta)

Galileo leverages an LLM to create natural language descriptions of all severe issues. It ingests session events to "watch" sessions and identify patterns in user behavior. It then asks itself questions about what is happening in those sessions and distills those answers into descriptions of the issue that is causing users to struggle.

This allows anyone -- regardless of technical experience -- to assess the impact on user experience and effectively prioritize a resolution.

Example of a standard JS error issue without natural language titles: TypeError: Cannot read properties of undefined (reading 'pc')

Natural language issues are currently in Beta. If you'd like access, you can activate them in the Issues tab, or by contacting us for a demo here.

📘

How Galileo AI Uses Data

LogRocket is committed to keeping all customer data safe and secure. The model used to calculate severity scores is proprietary and heavily redacted. It only uses generic counts of actions to keep data fully anonymous. Session data sent to LLMs is never used to train models, and the LLMs we use are fully compliant with SOC II, GDPR, and CCPA. All data sent is encrypted both in transit and at rest.

If you have additional questions, please reach out to your Customer Success Manager, or contact us at [email protected]