AI Features

Guide to Sublyzer's AI-powered features including Chat Analyzer, Auto-Patch, Insights, and Error Explanations.

AI-powered observability

Sublyzer uses AI throughout the platform to help you understand, analyze, and fix issues faster. From natural language chat to automatic bug fixes, AI makes observability accessible to everyone.

AI Features Overview

  • AI Chat Analyzer — Conversational interface for your data (PRO)
  • Auto-Patch — Automatic bug fixes via GitHub (PRO)
  • AI Insights — Automated recommendations
  • Error Explanations — Plain-language error descriptions
  • TestingAI — Autonomous browser testing agents (PRO)
AI Chat Analyzer (PRO)

The AI Chat Analyzer lets you have a conversation with your data. Ask questions in natural language and get instant answers backed by your actual metrics, errors, and performance data.

How to access

  1. Open any integration dashboard
  2. Click the AI chat icon in the bottom right corner
  3. Type your question and press Enter

Example questions

  • "What are my top 5 errors this week?"
  • "Why is my LCP score so high?"
  • "Compare my performance to last month"
  • "What's causing the error spike on Tuesday?"
  • "How can I improve my Core Web Vitals?"
  • "Explain error #12345 in simple terms"
  • "What are my most common user complaints?"
  • "Show me trends for the signup funnel"

Context awareness

The AI has full context about your integration including:

  • All errors with stack traces
  • Performance metrics and trends
  • User sessions and behavior
  • Security scan results
  • Historical data within your retention period
Auto-Patch Technology (PRO)

Auto-Patch uses AI to automatically analyze errors, understand your codebase, and generate fixes that are submitted as pull requests to your GitHub repository.

How it works

  1. Error is detected with stack trace
  2. AI analyzes the error context and your repository code
  3. AI generates a fix based on the error pattern
  4. Fix is committed to your repository (or suggested for review)

Configuration options

  • Mode: "Suggest fixes" (creates draft PRs) or "Auto-apply" (commits directly)
  • Severity threshold: Only patch errors above certain severity (low/medium/high/critical)
  • Max patches per day: Limit the number of automatic fixes (0-100)
  • Schedule: Realtime, hourly, or daily batch processing

Setup requirements

  1. Connect your GitHub account (Settings → Integrations → GitHub)
  2. Select the repository linked to your integration
  3. Configure Auto-Patch settings for your integration

For detailed setup, see: Auto-Patch Guide

AI Insights

AI Insights automatically analyzes your data and surfaces actionable recommendations without you having to ask. These appear in your dashboard and Health Score.

Types of insights

Performance Optimization

Suggestions to improve load times and Core Web Vitals

Error Patterns

Recurring errors that need attention

Security Recommendations

Vulnerabilities and how to fix them

User Experience

Friction points in user journeys

Cost Savings

Potential savings from optimizations

Best Practices

Industry standards you might be missing

Insight structure

Each insight includes:

  • Title — Brief description of the issue
  • Priority — High, Medium, or Low
  • Potential Savings — Estimated impact in dollars (where applicable)
  • Implementation Effort — Estimated complexity to fix
  • Recommended Action — Step-by-step fix guidance
Error explanations

Don't understand an error message? Sublyzer's AI explains errors in plain language, making debugging accessible even for less technical team members.

How to use

  1. Click on any error in your dashboard
  2. Click "Explain this error" button
  3. AI generates a plain-language explanation

Example explanation

TypeError: Cannot read properties of undefined (reading 'map')

AI Explanation: This error occurs when your code tries to use the .map() function on a variable that is undefined. This usually happens when data hasn't loaded yet or an API returned an unexpected empty response. To fix this, add a check like data?.map() or ensure the data exists before mapping.

TestingAI / Swarm AI (PRO)

TestingAI runs autonomous browser agents that test your application like real users. These AI-powered agents navigate your app, click buttons, fill forms, and report issues they find.

Features

  • Autonomous Navigation — Agents explore your app without scripts
  • Video Recording — Watch exactly what the agent did (MP4/WebM)
  • Accessibility Testing — Optional a11y checks during navigation
  • Authentication Support — Test protected pages with login credentials
  • Real-time Logs — Stream test progress via SSE

Configuration

  • Number of agents: 1-8 concurrent agents
  • Max pages: 1-20 pages per agent
  • Max clicks: 0-50 interactions per agent
  • Include accessibility: Enable/disable a11y testing

For detailed setup, see: TestingAI Guide

Note: TestingAI runs are limited to 1 per month per integration on PRO plans. FREE plans do not include TestingAI runs.

AI usage and limits

AI features usage varies by plan:

Feature

FREE

PRO

AI Chat Analyzer

Unlimited

Auto-Patch

Unlimited

AI Insights

Basic

Full

Error Explanations

Limited

Unlimited

TestingAI Runs

0/month

1/month

Best practices

Getting the most from AI

  • Be specific in chat questions for better answers
  • Review Auto-Patch PRs before merging in production
  • Use "Suggest fixes" mode first until you trust the AI
  • Check AI Insights weekly for new recommendations
  • Run TestingAI after major releases to catch regressions
  • Combine AI suggestions with human code review

Need help? Join our Discord.