How It Works

The three execution modes

ScrapeSpace has three ways to run a scraping task. Understanding them helps you get the most out of the platform.

1. Agent mode

When you type a prompt on the Agent page, ScrapeSpace spins up a cloud browser and an AI agent. The agent:

1. Reads your prompt
2. Takes a screenshot of the page
3. Analyzes the DOM to find interactive elements
4. Decides what action to take (click, type, scroll, run JavaScript)
5. Repeats until the data is extracted
6. Returns structured results

This is the most flexible mode — the agent can handle complex navigation, pagination, login flows, and dynamic content. It typically takes 30 seconds to a few minutes depending on complexity.

2. Replay mode (scraper refresh)

After a successful agent run, ScrapeSpace evaluates whether the task is repeatable. If it is, the agent's actions are saved as an automation — a deterministic sequence of browser actions.

Replays run the saved automation directly in the browser without any AI involvement. They're fast (typically 3–5 seconds) and use replay quota rather than AI run quota.

3. Heal mode (coming soon)

Websites change. When an automation replay fails because a selector moved or a page restructured, ScrapeSpace's self-healing system kicks in. It analyzes what broke and patches the automation automatically.

How automations work

1. You describe what data you need (an AI run creates the automation)
2. The automation can be replayed on-demand or on a schedule
3. Each replay runs the saved extraction — fast, deterministic, no AI needed
4. Your plan determines monthly AI run and replay budgets, schedule frequency, and (Free only) how many automations you can store at once