🧪 Table Studio is currently in Alpha. Turn it on from the Experimental features page (the toggle is per-user, so each teammate enables it on their own account).
Table Studio is a new, AI-first way to train a Browse AI robot. Instead of training the robot by clicking around a webpage, you describe the columns you want and an AI agent figures out where to find that data on the page and trains a robot to extract it on a schedule.
Think of it as describing the spreadsheet you wish you had, and Browse AI does the rest.
What is Table Studio?
Table Studio is a robot training method that flips the traditional workflow:
In Robot Studio, you point and click on the live webpage to teach the robot what to capture. The table is the result.
In Table Studio, you start with the table. You design the columns (names, descriptions, types) and the AI agent works backwards to figure out how to extract that data from the live page.
After you design your columns, you click Preview and the AI agent:
Loads the live page.
Identifies where each column you described lives on the page.
Returns a sample of the extracted data so you can verify it looks right.
Once you're happy, you click Save Robot and a robot is trained that runs extractions from that page or any similar page on that website on demand or on a schedule.
When to use Table Studio vs. Robot Studio
Use Table Studio when… | Use Robot Studio when… |
The data you want is visible on the page without needing to click, log in, search, or fill a form first. | You need the robot to log in, click through tabs, fill a search form, or navigate before reaching the data. |
You already know the exact columns and structure you want in your final spreadsheet. | You'd rather select data visually by clicking on the page. |
You want the AI agent to figure out the page structure for you. | The page has unusual structure where pointing and clicking is more reliable. |
💡 If you're not sure which one to use, try Table Studio first. It's the fastest path from "I want this data" to "I have a working robot", and you can always retrain with Robot Studio later if the page needs interaction.
How to train a new robot with Table Studio
Step 1: Start training
From your Browse AI dashboard, click Build New Robot.
Enter the Origin URL: the page that already shows the data you want.
Click Start Training Robot. A popover appears with two sections: AI Training (Table Studio) and Visual Training (Robot Studio / Chrome Extension).
Under AI Training, click Table Studio.
💡 Pick the right starting URL. The closer your Origin URL is to the page that already shows the data, the better Table Studio works. For example, point it at example.com/products, not example.com's homepage.
Step 2: Design your columns
Table Studio opens with one starter column. This is where you describe the spreadsheet you want.
For each column:
Name the column the way you'd want it to appear in your final spreadsheet (e.g. Product name, Price (USD), Posted date).
Add a short description if the name alone isn't obvious. The AI agent uses the description to decide which element on the page maps to your column.
Concrete descriptions help, especially when the page shows multiple similar values. For example, the listed price before any discount and the final price after discounts and taxes tell the agent which number to grab when both are visible.Add more columns as needed using + Add column. Columns can be plain text (the default), List for repeating items (see How to capture a list of items below), or Screenshot to capture an image of a region of the page.
Step 3: Preview the data
When you're happy with your columns, click Preview. The AI agent:
Loads the live page.
Tries to identify each column you described.
Returns a sample of the extracted data, right in your table.
Use the preview to sanity-check the AI agent's interpretation:
Did each column pull the right value?
Are numbers, dates, and currencies in the format you expected?
Are there any columns that came back empty or with the wrong content?
If something looks off, edit your columns and click Preview again. Each preview takes a minute or two while the AI agent loads the page and works through every column, so try to bundle your changes into one pass before previewing again. Your robot and any existing data stay untouched until you click Save Robot.
Step 4: Save your robot
When the preview row looks right, click Save Robot.
Browse AI will:
Train and save a robot based on the columns you designed, optimized for high accuracy and low cost, with your data schema and target website.
Kick off a fresh extraction against your Origin URL so the live Table starts populating with real data.
Show a status bar at the bottom of your Table while that first extraction runs.
After the first extraction completes, your robot behaves like any other Browse AI robot. You can set up monitoring, run bulk extractions, connect integrations, build workflows, or export the data.
📖 The first time you train a robot in Table Studio, you'll also see a short guided tour highlighting the most useful next steps: Deep scraping, Monitoring, Integrations, Workflows, and Calculated columns. You can dismiss it with the X if you'd rather explore on your own.
How to capture a list of items
Many pages mix one-off values (a total result count, a page title, a hero price) with a repeating list of items below them. Table Studio handles both at once, but they live in different tabs.
One-off values that appear once on the page go in the Main tab as regular columns. For example: Total results found, Page title, Currency.
Repeating items are captured by adding a column with the type set to List. When you add a List column, a new tab is added at the bottom of the screen. Open that tab and design the columns you want for each item inside the list (e.g., Job title, Company, Posted date).
You can capture multiple lists from the same page. Add one List column per list you care about. For example, on a search results page you might add one List column for the categories in the sidebar and another for the main search results. Each List column creates its own tab, so the two lists stay separate and clearly labeled.
💡 The list tabs can be easy to miss. After you add a List column, check the tab bar at the bottom of the screen. The list's columns are designed there, not in the Main tab.
When you press Preview, the AI agent fills in the Main tab values and the rows inside every list tab. Use the tabs together to verify the extraction looks right (see How does Table Studio handle pagination and long lists? below for how to check each list).
Adding more columns or lists to an existing table
You can add new columns, or whole new lists, to a table you've already trained, without recreating the robot from scratch. Open the table, switch it into Table Studio, design what you want to add, preview, and save.
Open the robot's Table.
Click + Add column next to the last column in the table, or click the ... icon on any column header and press Edit column.
Choose Open Design Mode to switch the table into Design Mode.
From there you can:
Add new text or Screenshot columns.
Add a List column to capture an extra list of items on the page (see How to capture a list of items above).
Rename or reword existing columns.
Click Preview to see one set of sample data extracted from your robot's default Origin URL.
Your robot and existing Table data stay untouched until you click Save Robot. Once saved, the robot is updated and a fresh extraction is queued so the new columns start populating with real data.
⚠️ If your existing robot was trained with clicks, inputs, or logins (for example, a robot that signs into a site or fills a search form), Table Studio can't edit it yet.
To add columns to that kind of robot, you will be asked to retrain it visually in Robot Studio.
Reusing the column structure from another Table Studio robot
If you've already trained one robot in Table Studio and want a second robot to produce the same shape of data (same columns), you can copy the schema instead of recreating it from scratch. This makes it possible to reliably scrape the same data schema from 100s or thousands of websites with very little time spent on setup.
When creating a new robot, check the Copy columns (schema) from existing table option.
Pick the existing table you want to copy from. Only tables that have been designed in Table Studio are listed.
Continue the normal Table Studio flow. You'll start with the same columns already in place.
💡 Only tables originally created in Table Studio can be used as a reference. Robots trained with Robot Studio or the Chrome extension don't appear in the picker.
Current limitations
Table Studio is in Alpha. There are a couple of things it doesn't do yet:
No interactions before extraction. Table Studio expects the data to already be visible on the page when it loads. It doesn't click buttons, fill in search boxes, or log in. For sites that need any of that, use Robot Studio.
Can't edit click/input/login robots. If a robot was trained with interactions, Design Mode can't add columns to it. You'll need to retrain it with Robot Studio.
Removing these limitations is on our product roadmap for Q3 2026.
Frequently asked questions
How is Table Studio different from a Robot Studio robot?
In Robot Studio, you teach the robot by clicking on the page; the columns come out of what you click. In Table Studio, you describe the columns first and the AI agent works backwards to find them on the page. Same end result (a structured Table), very different starting point. Table Studio generally makes it faster to create data scraping pipelines, especially if you need the same data structure from many websites.
Our early users included individuals creating and reviewing high quality data scraping pipelines from 200+ websites in ~2 weeks using Table Studio.
Does using Table Studio cost extra credits?
Not for normal usage. While Table Studio is in Alpha, training and preview runs are free. Once Table Studio is generally available, they'll stay free at the volume that covers what most teams need. If your training usage goes well beyond that, additional preview and training runs will use your regular task credits.
Once your robot is saved and runs on a schedule, it consumes credits like any other Browse AI robot.
How does Table Studio handle pagination and long lists?
The AI agent figures out pagination and infinite scroll on its own based on the page's layout. There's nothing to configure.
To check that it's working, run Preview and look at the list count shown on your List column in the Main tab. If the count is what the page actually shows (or the limit you'd expect for that robot), the agent has paginated correctly. If it looks low, refine your List column's description (for example, mention that the list spans multiple pages or scrolls) and Preview again.
For previewing purposes, list items are capped. You will see a notice next to the count if the cap has applied.
Do my teammates also see Table Studio after I enable it?
No. The experimental feature toggle is per-user. Each teammate who wants to use Table Studio to create new tables or edit existing tables needs to enable it on their own account from the Experimental features page. Robots trained in Table Studio, however, are available to your whole workspace just like any other robot.
Can I use Table Studio to edit a robot I built with Robot Studio?
Yes, as long as the existing robot doesn't rely on anything Table Studio can't handle yet (clicks, form inputs, navigation, or login). If your robot just extracts text or a list from a page that already shows the data, you can switch its table into Table Studio and add or change columns from there. For robots that need interactions, retrain them with Robot Studio for now.
Can I switch a Table Studio robot to Robot Studio later?
Yes. You can retrain any robot using Robot Studio at any time. The existing data in the robot's Table is preserved.
A column came back with the wrong data. What should I do?
Before clicking Save Robot, edit the column's name or description to be more specific about which element on the page you mean, then click Preview again. Most miss-mappings can be fixed in one or two iterations. If you're still stuck, please share your table URL with [email protected] so our team can take a look and assist you.
What's next?
Set up monitoring so your Table stays up to date automatically.
Run bulk extractions to point this same robot at thousands of URLs.
Add calculated columns to enrich the data your robot extracts.
Connect an integration to push the data into Google Sheets, Slack, Airtable, or your own systems.