> For the complete documentation index, see [llms.txt](https://examind.gitbook.io/v1/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://examind.gitbook.io/v1/feedback-machines/build-a-feedback-machine.md).

# Build a Feedback Machine

Create a Feedback Machine in four steps — from assignment instructions to a published machine students can use.

Building a Feedback Machine is a four-step process: describe the assignment, set up a rubric, let the system generate the evaluation, then configure and publish. The builder walks you through each step in order.

{% stepper %}
{% step %}

### Assignment Instructions

Paste or type your assignment instructions. Feedback Machines reads them to understand the work being evaluated — the kind of submission expected (text, `.docx`, or `.pdf`), whether writing quality matters, and whether it's a group assignment.
{% endstep %}

{% step %}

### Rubric

Give your Feedback Machine a grading rubric. You have two options:

* **Import a Rubric** — paste an existing rubric in any format, and the system structures it for you.
* **Create a Rubric** — let Feedback Machines generate one. Use the optional **Rubric Preferences** field to direct it: the scoring system you want, point values, the criteria to emphasize, or the number of categories. Then select **Generate Rubric**.

A rubric is made of **Parts** (the criteria) and, within each part, **Levels** (performance tiers) with **point values**.
{% endstep %}

{% step %}

### Review and Modify Machine

Feedback Machines analyzes your assignment and rubric and generates the **evaluation questions** it will use to assess submissions — each with its own **performance levels** and points. This is the machine's grading logic, expressed as questions the AI can answer about each submission.

The machine review page shows the generated parts, questions, and performance levels. From here you can:

* [**Modify the machine**](/v1/feedback-machines/build-a-feedback-machine/modifying-a-feedback-machine.md) — change questions, adjust points, add instructor notes, and more, just by asking in plain language.
* [**Audit for overlapping levels**](/v1/feedback-machines/build-a-feedback-machine/modifying-a-feedback-machine.md#audit-the-machine-for-overlapping-levels) — have the machine check each question for performance levels that could both fit the same work, and review its suggested fixes.
* **Resolve** [**content dependencies**](/v1/feedback-machines/build-a-feedback-machine/content-dependencies.md) — provide any external material a question refers to (like a case study or template) so the AI can evaluate it accurately.
* **Understand** [**how scoring works**](/v1/feedback-machines/build-a-feedback-machine/how-scoring-works.md) — so the points reflect how you grade.

You can refine anything here, before or after publishing.
{% endstep %}

{% step %}

### Configure & Publish

Name your machine and set how it behaves for students:

* **Formative Feedback View** — when students see improvement feedback (**Until approved**, **Always**, or **Never**), and whether it carries **score estimates**.
* **Summative Evaluation View** — when students see their final results (**After review & approval**, **Always**, or **Never**), and whether they can expand each rubric part for the **detailed evaluation breakdown**.
* **Machine details** — show how the machine evaluates always, only after a first submission, or never.
* **Sharing** — keep the machine **personal** (just you) or share it with a **class**.
* **Submission type** — how students submit: **DOCX**, **PDF**, **DOCX or PDF** (either format, so a submission can even mix the two), or **text** (copy & paste). The format decides what the machine can take into account — see [What a Feedback Machine Can See in Submissions](/v1/feedback-machines/build-a-feedback-machine/what-a-feedback-machine-can-see.md). For file uploads, also set **Files Per Submission** (up to three) — when more than one is allowed, all files are evaluated together.
* **Evaluation standard** — how the machine resolves work that sits on the boundary between two performance levels: **Strict** awards the lower level, **Generous** the higher. A second switch applies the Generous standard to final submissions only. See [How Scoring Works](/v1/feedback-machines/build-a-feedback-machine/how-scoring-works.md#the-evaluation-standard).

If the machine is linked to an EXAMIND assessment, you'll also set **Treat EXAMIND submissions as** — whether work arriving from that assessment is formative, summative, or both.

Publish when you're ready, and the machine becomes available for use. You can change any of these settings later.

{% hint style="info" %}
The two view settings are worth a moment's thought — they decide whether students get unlimited practice feedback, a single final grade, or both. [What Students See](/v1/feedback-machines/use/what-students-see.md) explains each option and how approval releases a final result.
{% endhint %}
{% endstep %}
{% endstepper %}

{% hint style="success" %}
**No rubric yet? You don't need one.** Use **Create a Rubric** with **Rubric Preferences** — describe the qualities that matter for your learning outcomes and Feedback Machines drafts a complete, criteria-based rubric you can review and adjust.
{% endhint %}

## Next step

[Use a Feedback Machine →](/v1/feedback-machines/use.md)