Inputs & Context
Every AI workflow begins with the data you feed it. Understanding how Aeye receives, parses, and holds context across a session is the foundation of building reliable automations.
What is an input?
Aninputis any data passed into an Aeye workflow at runtime: text, structured JSON, file content, API responses, or user-submitted forms. Inputs are the raw material your workflow acts on.
Aeye accepts inputs from three sources: direct API calls, connected triggers (webhooks, scheduled events, UI actions), and upstream workflow steps passing data forward.
- Text & Prompt: Natural language strings passed directly to the AI step. Supports template variables and dynamic injection.
- Structured Data: JSON objects, CSV rows, or key-value pairs. Aeye flattens nested structures for easy referencing.
- File & Media: PDFs, images, and documents uploaded directly or fetched from a URL at workflow execution time.
- Step Output: Data produced by a previous step — search results, transformed records, or extracted fields.
Context window
Context is the shared memory of a single workflow run. Each step can read and write to the context object usingctx.*references. Context is scoped to one execution — it resets between runs unless you explicitly persist it.
Usectx.user,ctx.session_id, or any custom key to pass data between steps without hardcoding values into each prompt.