Welcome to the AI Work Hub!

Our Conversation Analytics Bot is designed to turn your raw chat logs into actionable business intelligence in just a few clicks.

Whether you are analyzing 50 conversations or 5,000, this guide will show you how to format your data, choose your submission method, and understand the comprehensive reports you receive.

Step 1: Prepare Your Data

Your chat logs must be exported from your chatbot or support platform in JSON format. The system groups conversations by roomId. At a minimum, your data should include:

• RoomId: The unique ID for the conversation.
• From: Who is speaking (e.g., "user", "bot", "agent").
• Message: The actual text of the message.
• Timestamp: When the message was sent.

*Note: Don't have the technical resources to export or format your data into JSON? [Contact us about our Custom Managed Services], and our data engineering team will clean and format your logs for you!*

Step 2: Choose Your Submission Method

We support five main ways to submit your JSON conversation data depending on your workflow size and automation needs:

• Method 1: Direct Text Paste: Best for quick tests and small conversation snippets.
• Method 2: File Attachment: Best for one-off analyses of single medium-sized JSON files.
• Method 3: Cloud Sync (BYOS): Best for automated, bulk processing via Google Drive or Dropbox (delivers full multi-file report bundles back to your cloud folder).
• Method 4: Developer API: Best for large-scale enterprise automation, background processing, and real-time data pipelines.
• Method 5: Custom Managed Services (Done-For-You): Best for teams with limited technical resources, custom CRM exports, or no time to clean and format raw data.

Review the comparison table below to choose the method that best fits your needs:

Step 3: Choose Your Topic & Language Settings 

• AI Free Clustering (Default): The AI will organically discover the most common topics on its own.
• Predefined Topics: Tell the AI exactly what to look for (e.g., "Billing, Shipping, Tech Support"), and it will force conversations into those specific categories.
• Report Language Customization: Report Language Customization: By default, the AI will auto-detect the language used in your conversations and generate the report in that same language. However, you can command the bot to output the final report in a specific language (e.g., "English" or "Spanish").

   

  *Note: To ensure absolute accuracy, the AI will translate all descriptive summaries and data points into your chosen language, but will intentionally leave "evidence quotes" in the original language of the customer. 

Step 4: Understanding Your Results (The Output Files)

1. The Master Developer File (analysis_full_report.json):

This single "Parent JSON" file combines all the report data below (Executive Summary, Topic Clusters, Sentiment Analysis, Knowledge Gaps, Negative Conversations, Severe Cases, etc.) into one structured object. You can use our Conversation Intelligence Dashboard to visualize and chat with your data, or feed this file into your own system or dashboard for analysis.

2. The Master Excel Dashboard (analysis_report.xlsx):

The most important file for business leaders and managers. This is a beautifully formatted, 6-sheet Excel workbook that provides a complete overview of your bot or team's performance.

• Sheet 1 (Exec Summary): High-level metrics, overall sentiment percentages, and a plain-English summary of bot performance.
• Sheet 2 (Topic Clusters): Shows the most common topics, their volume, and average sentiment.
• Sheet 3 (Sentiment Analysis): Breaks down why customers were happy or angry.
• Sheet 4 (Knowledge Gaps): The most critical sheet. It lists exactly what your bot didn't know, provides exact verbatim quotes from the user as evidence, and suggests a fix.
• Sheet 5 (Bot Performance): Strengths, weaknesses, and resolution quality.
• Sheet 6 (Recommendations): Prioritized, actionable steps to improve your support.

3. The Full Conversation Report (conversation_report.csv & .json):

A line-by-line breakdown of every single conversation analyzed. Perfect for importing into your own BI tools (like Tableau or PowerBI).

• Key Fields Included: conversation_key, topic, topic_sentiment (how the user felt about their problem), bot_satisfaction (how they felt about the bot's help), resolved (Yes/No), message_count, and key_issue.

4. The Negative Conversations Report (negative_conversations.csv & .json):

This report automatically filters out the noise and only shows you conversations where the user was frustrated or the bot failed.

• Key Fields Included: topic_sentiment_reason (why the user was upset), bot_satisfaction_reason (where the bot failed), and bot_gap (what information was missing).

5. The Severe Cases Report (severe_cases.csv & .json):

This is your "Ground Truth" emergency report. It isolates conversations containing notable failures—such as AI hallucinations, completely ignored requests, or major compliance risks.

• Key Fields Included: bot_gap and evidence_quote. This report bypasses all AI summaries to show you the exact, unedited quotes of where things went severely wrong so you can fix them immediately.

Step 5: Visualize Your Data

Ready to explore your results interactively? Take your analysis_full_report.json file and head over to our Conversation Intelligence Dashboard to view your charts and chat with your data!