> ## Documentation Index > Fetch the complete documentation index at: https://docs.auxia.io/llms.txt > Use this file to discover all available pages before exploring further. # Agent and AI Insights > For a conceptual overview of Agent and AI Insights, see [Agent and AI Insights](/concepts/ai-insights) in Product Concepts. ## 7.1 Accessing Agent and AI Insights ### Navigation 1. Click **Agent and AI Insights** in the sidebar 2. The chat interface opens ### Permissions Required You need the **Agent External Use** permission to access Agent and AI Insights. Contact your admin if you can't access this feature. ### Interface Overview | Element | Description | | --------------- | ------------------------ | | **Chat Area** | Conversation history | | **Input Box** | Where you type questions | | **Send Button** | Submit your question | | **History** | Previous conversations | *** ## 7.2 Asking Effective Questions ### Question Types **Performance Questions:** * "What's my conversion rate this week?" * "How many impressions did we get yesterday?" * "Show me CTR for the last 30 days" **Comparison Questions:** * "Compare treatment A vs treatment B" * "How does this week compare to last week?" * "Which journey is performing best?" **Trend Questions:** * "Is engagement trending up or down?" * "Show me the daily conversion trend" * "What happened to purchases over the past month?" **Investigation Questions:** * "Why did clicks drop on Monday?" * "What's driving the conversion increase?" * "Which surface has the highest engagement?" ### Tips for Better Answers **Be Specific:** * Less effective: "How are things going?" * More effective: "What's the conversion rate for the Welcome Journey this week?" **Include Context:** * Mention specific journeys, treatments, or time periods * Reference goals by name **Ask Follow-Up Questions:** * Dig deeper into initial findings * Ask "why" to understand causes ### Example Conversations **Example 1: Performance Check** ``` You: How did our New User Onboarding journey perform this week? AI: The New User Onboarding journey had strong performance this week: - 45,230 impressions (up 12% from last week) - 2,156 clicks (CTR: 4.8%) - 312 conversions (conversion rate: 0.69%) - Uplift vs control: +23% Top performing treatment: "Welcome Modal v2" with 5.2% CTR. ``` **Example 2: Investigation** ``` You: Why did engagement drop yesterday? AI: Looking at yesterday's data, I see a 35% drop in engagement. Potential causes: 1. The "Holiday Promo Banner" was paused at 2pm 2. This treatment typically drives 40% of daily engagement 3. Other treatments maintained normal levels Recommendation: Review whether the pause was intentional, and consider reactivating if appropriate. ``` *** ## 7.3 Understanding Responses ### Response Format Agent and AI Insights presents information in various formats: **Summary Statistics:** ``` Total Conversions: 1,234 Conversion Rate: 3.2% Week-over-Week Change: +15% ``` **Tables:** | Treatment | Impressions | CTR | | --------- | ----------- | ---- | | Banner A | 10,000 | 4.5% | | Banner B | 8,500 | 5.2% | **Insights:** Key observations highlighted in natural language. **Recommendations:** Suggested actions based on data analysis. ### Interpreting Results **Positive Indicators:** * Uplift above baseline * Improving trends * Above-average CTR **Areas for Investigation:** * Declining metrics * Below-average performance * Unexpected changes *** ## 7.4 Common Use Cases ### Daily Check-In Start your day with quick questions: * "What happened overnight?" * "Any anomalies in the last 24 hours?" * "How are my key journeys performing today?" ### Journey Monitoring Track active journeys: * "How is the Holiday Journey doing?" * "What's the CTR on our new treatment?" * "Are we on track for our conversion goal?" ### Performance Investigation Dig into issues: * "Why did journey X underperform?" * "What's different about treatment A vs B?" * "Which audience segment converts best?" ### Report Preparation Gather data for reports: * "Summarize this month's performance" * "What were the top 5 treatments by conversions?" * "Give me weekly trends for Q4" ### Optimization Planning Get improvement ideas: * "Which treatments should I focus on improving?" * "Where are we losing users in the funnel?" * "What optimization opportunities do you see?" *** ## 7.5 Prompt Templates Use these templates as starting points: ### Performance Review ``` How did [journey/treatment name] perform in [time period]? ``` ### Comparison ``` Compare [item A] vs [item B] for [time period] ``` ### Trend Analysis ``` Show me the [metric] trend for [time period] ``` ### Investigation ``` Why did [metric] [increase/decrease] on [date/period]? ``` ### Top Performers ``` What were the top [N] [treatments/journeys] by [metric] in [time period]? ``` ### Recommendations ``` What optimization opportunities do you see for [journey/goal]? ``` *** ## 7.6 Best Practices ### Start Broad, Then Narrow 1. Begin with high-level questions 2. Identify areas of interest 3. Ask follow-up questions for details ### Be Clear About Time Periods Specify time ranges explicitly: * "Last 7 days" * "This month" * "December 1-15" ### Reference Specific Entities Use exact names when possible: * Treatment names * Journey names * Goal names ### Save Important Findings When you get valuable insights: * Screenshot or copy important findings * Use for presentations and reports * Share with team members ### Verify Unusual Results If something seems unexpected: * Ask clarifying questions * Cross-check with Analytics dashboards * Consider data timing/delays *** ## 7.7 Limitations ### What Agent and AI Insights Can Do * Answer questions about your data * Provide summary statistics * Identify trends and patterns * Offer optimization suggestions ### What Agent and AI Insights Cannot Do | Limitation | Alternative | | -------------------------- | ------------------------- | | Make changes to treatments | Use Treatment editing | | Create new journeys | Use Journey Portfolio | | Access real-time data | Data has processing delay | | Predict future outcomes | Use historical analysis | ### Data Freshness Agent and AI Insights uses processed data, which may have a delay: * Most metrics: 1-4 hours * Some aggregations: Up to 24 hours For real-time counts, check the main dashboards. *** ## 7.8 Troubleshooting ### "I don't have data for that" **Possible Causes:** * Time period is too recent (data delay) * Journey/treatment doesn't exist * You may not have access to that data **Try:** * Use a slightly older time period * Verify the name of the journey/treatment * Contact admin about data access ### Generic or Unhelpful Responses **Tips:** * Be more specific in your question * Include time period * Name specific treatments or journeys * Ask simpler, more focused questions ### Slow Responses **Normal Wait:** * Complex queries may take 10-30 seconds * Large time ranges take longer **If Too Slow:** * Try a shorter time period * Ask a simpler question first * Refresh and try again *** ## 7.9 Privacy and Data ### What Data Can Agent and AI Insights Access? * Your company's Console data * Treatment performance metrics * Aggregated user statistics ### What Agent and AI Insights Cannot Access * Individual user data (PII) * Other companies' data * System configuration details ### Data Security * All queries are processed securely * Results are scoped to your permissions * Conversation history is private to you *** ## 7.10 Advanced Features ### Multi-Turn Conversations Agent and AI Insights remembers context within a conversation: ``` You: How did Banner A perform last week? AI: Banner A had 10,000 impressions and 450 clicks (4.5% CTR). You: And this week? AI: This week, Banner A had 12,000 impressions and 600 clicks (5.0% CTR) - a 0.5 percentage point improvement. ``` ### Comparative Analysis Ask for direct comparisons: ``` You: Compare all my active treatments by CTR for the last 30 days AI: Here are your active treatments ranked by CTR: | Treatment | CTR | Impressions | |-----------|-----|-------------| | Modal B | 6.2% | 15,000 | | Banner A | 4.5% | 45,000 | | Push v2 | 3.8% | 28,000 | | *(additional rows omitted)* | | | ``` ### Anomaly Detection Ask about unusual patterns: ``` You: Were there any anomalies in the last week? AI: I detected the following anomalies: 1. Dec 15: Push notification CTR dropped 40% (7pm-11pm) 2. Dec 17: Conversion rate spiked 85% (likely due to promo) ``` *** ## Next Section Continue to [Section 8: QA & Testing](/guides/comprehensive/qa-testing) for quality assurance documentation.