If you've ever had to ask your data team every time you were curious about your data—and then wait days for the results—you can now cut that time down dramatically.
Notebook is an analysis tool that turns your data into charts when you ask the AI directly. Even if you don't know code like SQL or Python, just ask in your everyday words—like "How was last week's chat satisfaction?"—and the AI finds the data and organizes the answer for you.
Because your customer and chat data is already stored in Channel Talk, you can start analyzing the moment you open Notebook—no need to export and move data like with other analysis tools. And if you want to look at your company's revenue or product data too, you can connect external data like Google Sheets or BigQuery and compare everything on one screen.
Notebook is currently in CBT (Closed Beta Test). We'll keep enhancing it and plan to release it officially.
Example 1. Preparing weekly executive reports Every Monday morning, type "Show me last week's chat resolution rate, average response time, satisfaction, and reopen rate" into Notebook, and the AI builds the charts automatically. Save the finished analysis as a dashboard, share it with your team via link, and it refreshes automatically with the latest data the following week.
Example 2. Analyzing promotion performance Ahead of a promotion, ask "Predict how chat volume will change compared to the same period last year," and the AI charts the trend for you. Follow up with questions like "Show me the top 5 topics with rising complaints" to dig deeper into the analysis.
Path: [Channel Settings → Data Sources & Analytics → Data Sources] or [Team → Notebook → Data Sources]
To start analyzing in Notebook, you first need to connect the data source you'll use. Channel Talk's internal data is ready to use with no setup, and external data only requires a one-time connection setup.
Data source | Description | Setup |
|---|---|---|
Channel Talk internal data | All chats, customer info, statistics, and agent performance | Ready to use |
PostgreSQL | Open-source relational database | One-time connection setup |
MySQL | Open-source relational database | One-time connection setup |
BigQuery | Google Cloud serverless data warehouse | One-time connection setup |
StarRocks | Real-time analytics data warehouse | One-time connection setup |
Google Sheets | Google Sheets | One-time connection setup |
You can preview connected data to see which tables it contains. When connection details change, recheck them with [Edit → Test Connection]. Note that data referenced by another Notebook can't be disconnected.
Path: [Team → Notebook → + Add Notebook]
1) Build it through conversation with AI CoS [Coming soon] You don't have to write SQL yourself. Just ask the AI Agent CoS in natural language, and it handles everything from retrieving data to building charts. You can choose between two modes depending on the situation.
Auto-approve mode: The AI runs all the way through without checking in. This is good for quickly drawing up a dashboard when you first create a Notebook.
Apply after review mode: The AI sends an approval request whenever it modifies or creates data, or accesses external data. This lets you review changes one by one when editing an existing Notebook.
2) Build it manually
Creating cells A Notebook is built from connected blocks called cells. Each cell serves a purpose—retrieving data, analysis, visualization, or explanation—and stacking multiple cells completes a single analysis report. Add a cell by clicking the [+] button in the bottom input field, or by clicking the button for the cell type you want.
Purpose | Cell type | Role |
|---|---|---|
Analysis | SQL | A cell that pulls the data you want directly. View results as a table right away, or visualize them as a chart through a [Chart] cell. |
Analysis | Python | Used for complex calculations or analysis that are hard to express with SQL alone. |
Analysis | File (coming soon) | Upload CSV, Excel, and other files needed for analysis. |
Visualization | Chart | A cell that draws your analysis results as a graph. Choose from bar, line, pie, heatmap, scatter, and area charts, set which values go on the horizontal and vertical axes, and decide how to group them—by sum, average, or count. |
Visualization | Table | A cell that organizes data into a table. It shows 25 rows per page, and you can page through to view all the data. |
Visualization | Markdown | A cell that adds written explanations between charts. Leaving notes like "why this analysis was done" or "what stands out" helps others understand the analysis easily. |
Condition | Inputs | A cell that separates out the conditions used in analysis for easier management. For example, if you set a date range as an Input, you can rerun the same analysis with just a different date—no code changes needed. |
Running cells Once you've created your cells, click [Run All] or [Run] on each cell to analyze the data or draw the charts.
Fixing errors When you run a cell, any cell with a problem will show an error message. Check the error and fix it.
Click the [Notebook/Dashboard] button in the top right to add and edit the cells you're working on, or preview it in report form.
Notebook: The screen where you build your analysis by adding and editing cells. Use it when asking the AI questions or exploring your data.
Dashboard: A screen that neatly collects only the finished charts and tables. The code and analysis process are hidden and only the results are shown, making it ideal for sharing with your team or executives. In the dashboard, you can drag each cell to adjust its position and size. Feel free to arrange your report however you like.
Once you've finished your analysis, share your Notebook with your team using [Publish].
Because a Notebook can contain sensitive data along with analysis results, it's set to Private by default when created. When you publish, you can set the permissions for what people can do when they join via an individual Notebook's link.
Share link options:
Private (only you): Default
All Managers in the channel: Grant either Edit or Read access
If you want to refresh the same data daily or weekly, you can set a schedule (automatic run) on a published Notebook. At the time you specify, the Notebook runs on its own, keeping your data always up to date.
Set the cycle: Specify the day of the week and time to run, on a daily or weekly basis.
Auto-refresh: When the scheduled time arrives, all cells in the Notebook rerun with the latest data.
Manage schedules: You can edit the time or cycle of a registered schedule, or delete it, anytime from the list.
Schedules can only be set on published Notebooks. You can't schedule a Notebook that hasn't been published yet, so publish it first and then try.