Manual Runs

Overview

A manual run is how you kick a bot off yourself, usually while you are iterating. You will do this constantly while you are building a new bot, and occasionally after that when you want an on-demand result outside the bot’s schedule.

Running a bot

From the bot page, click Run. The run dialog lets you optionally:

• Provide input data, a block of text or JSON that gets passed into the bot’s first stage.
• Upload files, which are made available to the bot during the run.

If you do not need either, you can run the bot as-is.

Input data

Input data is how you parameterise a run without rewriting tasks. If the same bot should work for different topics or different targets, put the topic or target in input data. The tasks can reference it in their prompts.

For example, a competitor-price-check bot might accept input data like:

{
  "competitor": "Acme Co",
  "product_page_url": "https://acme.example.com/pricing"
}

Each task can then refer to the competitor and URL directly.

File uploads

If your bot needs to work with a document, CSV, or image, attach it as a file on the run. The file is available for the duration of that run only.

Credits and cost

Each active task in a bot costs 50 credits per run. So a 1-task bot costs 50 credits per run, a 5-task bot costs 250 credits per run, and so on. The full cost is charged upfront when the run starts. If the run fails, the credits are refunded automatically. See Run History for how to tell the difference between a completed, failed, or refunded run.

What happens next

Once you kick off the run, it moves through a lifecycle: queued, running, completed (or failed). You can watch it progress from the run page. When it is done, open the run to see the output, per-task outputs, and any errors.

When you are happy with the output from a few manual runs, switch to a schedule so the bot runs itself.