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

# Classifier

> Classify text using AI into categories you define. Great for tagging feedback, categorizing content, labeling intents, or sorting leads by topic or sentiment.

## What It Does

* **Classifies text using AI** into categories (or “keys”) that you define.
* **Supports single or multi-label classification**, depending on your settings.
* **Optional reasoning** helps explain why the label was chosen.
* **Fully configurable** with custom keys, instructions, and optional column naming.
* **Auto-resolves naming conflicts** so your original data stays safe.

***

## 🏁 Getting Started

<Frame>
  <img src="https://mintcdn.com/nurturev/OitEIaKlfl7lCKCJ/images/Classifier%20node%20configuration%20screenshot.png?fit=max&auto=format&n=OitEIaKlfl7lCKCJ&q=85&s=9195c0d84ccf544b791e64f073aca084" alt="Classifier node configuration screenshot" style={{ borderRadius: '0.5rem', width: '100%', margin: '1.5rem 0' }} width="1182" height="1664" data-path="images/Classifier node configuration screenshot.png" />
</Frame>

<Steps>
  <Step title="Add the Node">
    Drop the **Classifier** node anywhere in your playbook.
  </Step>

  <Step title="Choose Input Column">
    Select the column that contains the text to classify. Use **Insert Input** or type `@` to reference earlier steps.
  </Step>

  <Step title="Write Your Instructions">
    Optionally, guide the model with a prompt like “Classify this as pricing, support, or product.”
  </Step>

  <Step title="Define Keys">
    Add labels for the AI to pick from. Each key can include:

    * A name (e.g. `support`)
    * A description to help improve accuracy
  </Step>

  <Step title="Set Preferences">
    * Toggle **Include Reason** to explain why the label was picked
    * Toggle **Allow Multiple Classes** to enable multi-label tagging
    * Optionally rename the output column
  </Step>

  <Step title="Run the Node">
    Each row will now return a classification result (and optionally a reason).
  </Step>
</Steps>

***

## Inputs

### 🛠️ Required Fields

* **Text Column (✅)**\
  Select the column you want to classify — e.g., `@feedback`, `@summary`, etc.

* **Classification Keys (✅)**\
  Define the labels you want applied. Each key includes:
  * A name (e.g. `product`, `pricing`)
  * An optional description (e.g. “Mentions product bugs or roadmap features”)

### 🎯 Optional Fields

* **Instructions (⚪️)**\
  Helps the AI understand how to apply your keys. Example: “Tag this as product, pricing, or support.”

* **Include Reason (⚪️)**\
  Adds a second column with short explanations.\
  *Adds 1 extra credit per row.*

* **Allow Multiple Classes (⚪️)**\
  Enables multiple labels per row.\
  *Returns values as arrays like `["product", "pricing"]`.*

* **Output Column Name (⚪️)**\
  Customize the result column. Defaults to `CLASS`.\
  Auto-suffixed if there’s a conflict (`CLASS_1`, etc).

***

## Output

The node adds:

| Output Column           | Description                                    |
| ----------------------- | ---------------------------------------------- |
| `CLASS` (default)       | The predicted label(s).                        |
| `CLASSIFICATION_REASON` | (Optional) Explains why each label was chosen. |

<Note>
  If multi-label is on, values appear as arrays like `["support", "pricing"]`.
</Note>

***

## 🚀 Example Use Cases

| Use Case               | Setup Example                                                      |
| ---------------------- | ------------------------------------------------------------------ |
| Feedback Tagging       | Classify `@feedback` into `product`, `pricing`, `support`, `other` |
| Sentiment Analysis     | Classify `@review_text` as `positive`, `neutral`, `negative`       |
| Intent Detection       | Tag `@chat_log` for `buying intent`, `support`, `info-seeking`     |
| Content Classification | Label `@summary` into `news`, `promo`, `update`, `other`           |

***

## ✨ Pro Tips

<Tip>
  Use short **descriptions** for each key — it makes classification more accurate and predictable.
</Tip>

<Tip>
  Add fields using `@` or the **Insert Input** button — avoids typos and ensures clean references.
</Tip>

<Tip>
  Turn on **Allow Multiple Classes** when text may touch multiple themes.
</Tip>

<Tip>
  Use a custom output name like `tag` or `intent_label` to make downstream logic clearer.
</Tip>

***

## ⚠️ Important Considerations

<Warning>
  Multi-class output returns **arrays** like `["product", "support"]`. Plan downstream filters accordingly.
</Warning>

<Warning>
  **Include Reason** increases cost by +1 credit per row.
</Warning>

***

## 🛠 Troubleshooting & Gotchas

| Symptom                    | Likely Cause            | Fix                                           |
| -------------------------- | ----------------------- | --------------------------------------------- |
| Empty result column        | No keys provided        | Add at least one classification key           |
| Same label for all rows    | Instructions too vague  | Add clearer rules or improve key descriptions |
| `CLASS_1`, `CLASS_2`, etc. | Name conflict in output | Rename the output column in config            |

***

## 📝 FAQ

<AccordionGroup>
  <Accordion title="Can I use multiple labels per row?">
    Yes — toggle **Allow Multiple Classes** to get array outputs like `["pricing", "product"]`.
  </Accordion>

  <Accordion title="What does 'Include Reason' do?">
    Adds a second column explaining why the label was chosen. Great for review or debugging.
  </Accordion>

  <Accordion title="How many labels can I use?">
    There’s no hard limit, but 3–10 keys works best for clarity and model accuracy.
  </Accordion>
</AccordionGroup>

***

## 💰 Pricing

| Mode                | Credit Cost per Row |
| ------------------- | ------------------- |
| Classification only | 1 credit            |
| With Reason         | 2 credits           |

<Note>
  Only rows that receive a result are charged. Enabling **Include Reason** adds +1 credit per row.
</Note>

***

<p style={{ fontSize: '1rem', fontWeight: 'bold', marginTop: '1.5rem' }}>
  Classify anything — feedback, intent, topic, or theme — and make your data more actionable with just a few clicks. 🏷️⚡
</p>