# Chatting with Your Data

The AI chat in Weavestream is a conversational interface for exploring and analyzing your data. You ask questions in natural language, and the AI responds with insights drawn from your actual items.

# How It Works

When you send a message, Weavestream doesn't just forward your question to the AI. It goes through a smart process:

1. Smart Routing — The AI first analyzes your question to figure out which fields and data are most relevant. This prevents sending unnecessary data and keeps responses focused.
2. Context Building — Weavestream assembles the relevant items and fields, including sample values so the AI understands the data structure.
3. Analysis — Your question and the relevant data are sent to the AI provider, which returns its analysis.
4. Display — The response appears in the chat with markdown formatting (bold, italic, code, lists, etc.).

# What Data Does the AI See?

The AI's scope depends on what you've selected in the sidebar:

• A specific endpoint — The AI sees items from that endpoint
• A source — The AI sees items from all endpoints in that source
• A Smart Filter — The AI sees items matching the filter's rules
• All Items — The AI sees items from all sources

A pill indicator at the bottom of the chat shows what's currently selected (e.g., "Open Alerts" or "All Items").

If your Smart Filter has Output Fields configured, only those fields are sent to the AI. This is a great way to focus the analysis and reduce noise.

# Asking Good Questions

The AI works best with specific, clear questions. Here are some examples:

Summaries and overviews:

• "Give me a summary of the most important items from the last week"
• "What are the top 5 issues by severity?"

Pattern detection:

• "Are there any recurring patterns in these alerts?"
• "Which devices have the most failures?"

Comparisons:

• "Compare this week's data to last week"
• "Which endpoints have the highest error rates?"

Action-oriented:

• "What should I prioritize fixing first?"
• "Create a list of action items based on these tickets"

Specific lookups:

• "Find all items related to authentication failures"
• "Which items were created by user X?"

# Conversations

# Following Up

The AI remembers the context of your conversation. You can ask follow-up questions without repeating yourself:

1. "Summarize the critical alerts"
2. "Which of those have been open the longest?"
3. "What would you recommend I do about the top 3?"

# Managing Conversations

You can have multiple saved conversations. Use the conversation selector dropdown at the top of the chat to:

• Switch between conversations — Click a conversation name to load it
• Rename a conversation — Hover over a conversation and use the rename option
• Delete a conversation — Remove conversations you no longer need
• Start a new chat — Click the new chat button (pencil icon) to begin fresh

Conversations are saved automatically and persist between sessions.

# Smart Chat Naming

If Apple Intelligence is available on your Mac, Weavestream can automatically generate descriptive names for your conversations based on the content. This saves you from having to manually rename each chat.

# The Info Button

After an AI response, you'll see a small info button (i) you can click to see details about the analysis:

• Items analyzed — How many items were sent to the AI out of the total available (e.g., "Analyzed 42 of 156 items")
• Token usage — Information about how much of the AI's context was used

This is helpful for understanding whether the AI saw all your data or just a subset.

# Tips

• Select the right scope. If you want the AI to analyze a specific endpoint, select it in the sidebar before asking your question. Selecting "All Items" with thousands of items may result in only a subset being analyzed.
• Use Smart Filters with Output Fields. This gives you the most focused, relevant analysis.
• Start broad, then narrow down. Ask for a summary first, then drill into specific areas with follow-up questions.
• The AI can search. Ask "find all items where X is Y" and the AI will look through the data for matches.

# Next Steps

• Custom Analysis Prompts — Create reusable analysis templates
• Privacy & AI Providers — Understand data handling