> ## Documentation Index
> Fetch the complete documentation index at: https://docs.allyy.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Workflows

> Automate the data → model → prediction → export loop on a schedule

Doing the loop once by hand is fine for a pilot. For live programmes you want it to run on its own. **Workflows** chain the steps together and run them on a schedule, so a fresh, optimised decision is ready whenever you need it.

## What a workflow can chain

> [card] Sync data
>
> Pull the latest from your batch data sources before anything else runs.

> [card] Train / retrain
>
> Refresh a model on the newest data so it doesn't drift.

> [card] Generate predictions
>
> Score the current population with the up-to-date model.

> [card] Export
>
> Deliver the resulting decision to your connected tools.

## A typical recurring campaign

**1. Sync the latest supporter & gift data**

**2. Retrain the model so it reflects recent behaviour**

**3. Score the eligible population**

**4. Produce the optimised decision**

**5. Export it to your campaign tool**

Scheduled to run before each appeal, this means your team always works from a current, optimised list — with no manual clicks.

> [!NOTE]
>
> **Screenshot to add** — the workflow builder showing a chained sync → train → score → export pipeline and its schedule.

> [!NOTE]
>
> **Streaming** data sources flow continuously and aren't scheduled; workflows are for **batch** steps. See [Connect a data source](https://docs.allyy.io/processes/data.md) for the batch vs streaming distinction.

## Keeping an eye on it

Every workflow run — and each step within it — is tracked on the [Monitoring](https://docs.allyy.io/processes/monitoring_dashboard.md) page, colour-coded by success, in-progress or failure, so you can confirm the pipeline ran and catch problems early.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.