Skip to content
Loggator

AI Chat Assistant

The AI Chat Assistant is Loggator’s most powerful feature, enabling you to analyze logs using natural language instead of complex queries.

Overview

Section titled “Overview”

Chat with an AI assistant that can:

  • Search logs using natural language
  • Analyze container health automatically
  • Explain errors in plain language
  • Suggest solutions to problems
  • Filter and aggregate log data
  • Answer questions about your infrastructure

Getting Started

Section titled “Getting Started”

Enable the AI Assistant

Section titled “Enable the AI Assistant”
  1. Get an OpenRouter API key (free tier available)
  2. Add to your .env file:
    Terminal window
    OPENROUTER_API_KEY=sk-or-v1-your-key-here
    AI_MODEL=xiaomi/mimo-v2-flash:free
  3. Restart Loggator: docker compose restart loggator

Access the Chat

Section titled “Access the Chat”

Navigate to http://localhost:3000 and find the chat interface in the sidebar or main page.

Example Queries

Section titled “Example Queries”

Basic Questions

Section titled “Basic Questions”
What containers are you monitoring?
Show me recent logs
What's happening right now?

Response: List of monitored containers with their status

Error Detection

Section titled “Error Detection”
Show me all errors
Find errors from the last hour
What errors does nginx have?
Are there any exceptions?

Response: Filtered error logs with analysis

Container-Specific

Section titled “Container-Specific”
Tell me about the api container
Is my database healthy?
Show logs from the web container
What's the status of nginx?

Response: Container info and relevant logs

Time-Based Analysis

Section titled “Time-Based Analysis”
What happened in the last 30 minutes?
Show errors from the last 2 hours
Any warnings today?
What changed since yesterday?

Response: Time-filtered logs with context

Health Checks

Section titled “Health Checks”
Is everything running smoothly?
Check the health of all containers
Are there any problems?
Analyze the api container

Response: Health analysis with suggestions

Troubleshooting

Section titled “Troubleshooting”
Why is my api container failing?
What caused the recent errors?
Explain this error: [paste error]
How do I fix connection timeouts?

Response: Error explanation with solutions

Pattern Analysis

Section titled “Pattern Analysis”
Find all timeout errors
Show me authentication failures
Are there any memory issues?
What 500 errors occurred?

Response: Pattern-matched logs with analysis

AI Capabilities

Section titled “AI Capabilities”

Automatic Tool Calling

Section titled “Automatic Tool Calling”

The AI automatically uses these tools based on your query:

1. search_logs

Section titled “1. search_logs”

Searches logs with filters

Query: "Show me errors from nginx"
Tool: search_logs(query="error", container="nginx")

2. list_containers

Section titled “2. list_containers”

Lists monitored containers

Query: "What containers are running?"
Tool: list_containers(all=false)

3. get_container_info

Section titled “3. get_container_info”

Gets detailed container information

Query: "Tell me about the database"
Tool: get_container_info(containerName="database")

4. analyze_container_health

Section titled “4. analyze_container_health”

Analyzes container health

Query: "Is the api healthy?"
Tool: analyze_container_health(containerName="api", minutes=60)

Context Understanding

Section titled “Context Understanding”

The AI maintains context across conversations:

You: Show me nginx logs
AI: [Shows nginx logs]


You: Any errors in there?
AI: [Filters for errors in nginx logs - understands "there" = nginx]


You: What about timeouts?
AI: [Further filters nginx errors for timeout-related entries]

Response Formatting

Section titled “Response Formatting”

Responses include:

  • Markdown formatting for readability
  • Syntax-highlighted code blocks for log entries
  • Tables for structured data
  • Lists for multiple items
  • Timestamps in human-readable format

Advanced Usage

Section titled “Advanced Usage”

Complex Queries

Section titled “Complex Queries”

Combine multiple filters:

Show me all 500 errors from the api container
logged to stderr in the last 2 hours

The AI will:

  1. Parse your requirements
  2. Call search_logs with appropriate filters
  3. Analyze the results
  4. Provide formatted output with insights

Multi-Step Analysis

Section titled “Multi-Step Analysis”

Ask follow-up questions:

You: What containers have errors?
AI: [Lists containers with error counts]


You: Show me the errors from the top one
AI: [Shows errors from the container with most errors]


You: What's causing those errors?
AI: [Analyzes error patterns and suggests causes]


You: How do I fix it?
AI: [Provides solutions based on error types]

Aggregation and Statistics

Section titled “Aggregation and Statistics”
How many errors per container?
What's the error rate over time?
Which container logs the most?
Show me a summary of the last hour

Comparative Analysis

Section titled “Comparative Analysis”
Compare logs from api and database
Did errors increase in the last hour?
Which containers are noisiest?

Chat Features

Section titled “Chat Features”

Persistent History

Section titled “Persistent History”
  • Conversations are saved in your browser
  • Context is maintained across page reloads
  • Clear history button to start fresh

Tool Visibility

Section titled “Tool Visibility”

Watch the AI work:

  • See which tools are being called
  • View parameters passed to tools
  • Inspect raw tool results

Toggle visibility in the chat settings.

Regenerate Responses

Section titled “Regenerate Responses”

Don’t like a response? Request regeneration:

  • Click the “Regenerate” button
  • AI will retry with different reasoning
  • Useful for getting more detail or different perspective

Copy to Clipboard

Section titled “Copy to Clipboard”

Every response has a copy button:

  • Copy formatted markdown
  • Share with team members
  • Save to documentation

Best Practices

Section titled “Best Practices”

1. Be Specific

Section titled “1. Be Specific”

❌ Vague: “Show me logs” ✅ Specific: “Show me error logs from nginx in the last hour”

2. Use Natural Language

Section titled “2. Use Natural Language”

Don’t try to write code: ❌ search(query="error", filter="container=nginx") ✅ “Show me nginx errors”

3. Ask Follow-Up Questions

Section titled “3. Ask Follow-Up Questions”

Build on previous context:

1. What containers are monitored?
2. Show logs from the first one
3. Any errors there?
4. What caused them?

4. Be Patient

Section titled “4. Be Patient”

AI tools can take a few seconds:

  • Searching large log sets
  • Analyzing patterns
  • Generating explanations

5. Provide Context

Section titled “5. Provide Context”

Help the AI understand: ❌ “Fix it” ✅ “How do I fix the connection timeout error in my api container?”

Example Workflows

Section titled “Example Workflows”

Morning Health Check

Section titled “Morning Health Check”
1. Is everything running smoothly?
2. Show me any errors from the last 8 hours
3. Are there any unusual patterns?
4. Check the health of all containers

Incident Investigation

Section titled “Incident Investigation”
1. What happened at 14:30?
2. Show me errors around that time
3. Which containers were affected?
4. What was the root cause?
5. How can I prevent this?

Performance Analysis

Section titled “Performance Analysis”
1. Show me all timeout errors today
2. Which container has the most timeouts?
3. Analyze that container's health
4. What's causing the timeouts?

Deployment Verification

Section titled “Deployment Verification”
1. Show logs from the last 5 minutes
2. Are there any errors after deployment?
3. Is the new version healthy?
4. Compare with previous hour

Limitations

Section titled “Limitations”

Log Volume

Section titled “Log Volume”
  • AI analyzes up to 100 logs per query (default)
  • Large result sets are summarized
  • Use time filters for better results

Rate Limits

Section titled “Rate Limits”

Free models have limits:

  • 10-20 requests per minute
  • 100-200 requests per day

See AI Configuration for details.

Context Window

Section titled “Context Window”

AI has limited memory:

  • Keeps recent conversation context
  • May forget older messages
  • Start fresh for new topics

Model Capabilities

Section titled “Model Capabilities”

Different models have different strengths:

  • Free models: Good for basic queries
  • Paid models: Better reasoning and explanations

Troubleshooting

Section titled “Troubleshooting”

”AI features disabled”

Section titled “”AI features disabled””

Cause: OpenRouter API key not configured

Solution:

Terminal window
# Add to .env
OPENROUTER_API_KEY=sk-or-v1-your-key-here


# Restart
docker compose restart loggator

Slow responses

Section titled “Slow responses”

Causes:

  • Large result sets
  • Complex queries
  • Slow AI model

Solutions:

  1. Be more specific (reduces search scope)
  2. Use time filters
  3. Switch to a faster model (e.g., xiaomi/mimo-v2-flash:free)

Incorrect responses

Section titled “Incorrect responses”

Solutions:

  1. Rephrase your question
  2. Be more specific
  3. Try a different model
  4. Check if logs exist (use search page)

Rate limit errors

Section titled “Rate limit errors”

Solutions:

  1. Wait for limit to reset (usually 1 minute)
  2. Use direct search for simple queries
  3. Upgrade to paid model

Security & Privacy

Section titled “Security & Privacy”

What Data is Sent?

Section titled “What Data is Sent?”

Only log data matching your query:

  • Log messages
  • Timestamps
  • Container names
  • Metadata

Data Usage

Section titled “Data Usage”
  • Processed by OpenRouter’s AI models
  • May be used for model improvement (check provider policy)
  • Not stored by Loggator

Best Practices

Section titled “Best Practices”
  1. Don’t log sensitive data in containers
  2. Use label filtering to exclude sensitive containers
  3. Review queries before sending
  4. Rotate API keys regularly

See Security Considerations for more.

Next Steps

Section titled “Next Steps”