> ## Documentation Index
> Fetch the complete documentation index at: https://browseract-mintlify-update-docs-navigation-33744.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.
# Core Concepts
> In BrowserAct, everything comes from three ideas: workflows, nodes, and runs.
Before you dive into BrowserAct, we’ll quickly introduce three key concepts: **workflow**, **node**, and **task**.
Any time you want to turn a repetitive web operation (e.g. checking prices, scraping data, syncing lists) into an automated process in BrowserAct, you’ll be using these three:
* **Workflow** – the end-to-end process designed to achieve a specific business goal.
* **Node** – one small concrete step inside that process.
* **Task** – one actual run of that process, including its inputs, execution, and results.
In the rest of this section, we’ll look at what each of them means and how they map to the way you already work in a browser.
***
## 1. What is a workflow?
> **Workflow = the complete sequence of steps that lets BrowserAct follow your instructions and “do the work” on web pages for you.**
From a business perspective:
* Each workflow represents a **clear business goal**, such as “scrape Google News”, “collect TikTok hashtag videos”, or “get Google Maps business listings”.
* Every card you see in the **Templates** gallery (e.g. *YouTube Comments Scraper*, *Google Maps Scraper*) is essentially a pre-built workflow.
### From manual work to a workflow
Suppose you do this manually every day:
> Open an e-commerce site → search for “shoes” → go through all pages → copy product name, price, and link → paste into a spreadsheet
Seen as a workflow, you only keep the key steps:
1. Open the product listing page
2. Search by a keyword (for example, “shoes”)
3. Go through all listing pages
4. Collect each product’s name, price, and link
5. Save the collected data into a table
In BrowserAct, you turn these steps into one complete workflow, for example:
> **“E-commerce Product Price Scraper” workflow**
After that you can:
* Use the **same workflow** with different inputs (for example, different keywords or start URLs)
* Click **Run**, and BrowserAct will follow this route automatically to finish the job for you
**In short:**
* From the system’s point of view, a workflow is a set of nodes connected in order.
* From the business point of view, a workflow represents **the business goal itself** that you want to achieve.
***
## 2. What is a node?
Once you have a complete workflow, you can look one level deeper.
> **Node = the smallest execution unit inside a workflow — a single instruction that tells the browser to do one specific thing.**
Using the same example:\
At the workflow level, you might have a step called:
> “Search by keyword”
When this actually runs, it must be broken down into smaller instructions:
* Fill the keyword into the search input box
* Click the search button
In BrowserAct, these become **multiple nodes**, not just one.
### Manual actions vs. nodes
Here’s how common manual actions map to node types:
| Manual action | Typical node in BrowserAct |
| :---------------------------------- | :------------------------- |
| Open a URL | **Visit Page** node |
| Type text into a search box | **Input Text** node |
| Click a “Search” button | **Click Element** node |
| Scroll down the page | **Scroll Page** node |
| Click “Next page” | **Pagination** node |
| Read title, price, or other fields | **Extract Data** node |
| Repeat the same steps for each item | **Loop** node |
On the workflow canvas, each small block you see is one node.
So you can think of it like this:
* The **workflow** describes which steps are needed to reach a business goal (search → browse pages → extract data → save result).
* **Nodes** break these steps down into browser-level actions (which input to fill, which button to click, how to loop, which fields to extract, etc.).
**In short:**
> A workflow explains **how the whole thing goes from A to B**.\
> Nodes define **what exactly happens at each step on the web page**.
***
## 3. What is a task?
Now we have:
* A **workflow** – a complete process design (business goal + a set of nodes)
* **Nodes** – the individual instructions inside that process
Next comes the **task**.
> ***When you click “Run” on a workflow and it executes from start to finish, that single run is called a task.***
Each task has its own:
* **Input**
* For example: is the keyword “shoes” or “hats”? What is the start URL?
* **Execution details**
* Which nodes were run, in what order, and where (if anywhere) it failed
* **Result**
* The data collected
* Logs and status (success / failure)
* **Resource usage**
* How many credits this run consumed
The same workflow can generate many tasks. For example:
* Run once today to **scrape this week’s new products** → Task #1
* Run again tomorrow on the same workflow to **scrape next week’s new products** → Task #2
**In short:**
> A **task** is one concrete execution of a workflow at a specific time, with specific inputs, and its full record of what happened and what was produced.
***
You can now go to the **Template** **Marketplace**, pick any workflow, and use this mental model to see:
* Which business goal the workflow represents
* How it is built from nodes
* And what tasks are created each time you run it.