> For the complete documentation index, see [llms.txt](https://docs.nected.ai/nected-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.nected.ai/nected-docs/workflow/add-node/action-nodes/ai-agent-node/text-analysis.md).

# Text Analysis

The **Text Analysis Node** allows you to perform natural language understanding directly within your workflows — without external scripts or APIs.\
You can use it to analyze text, extract key insights, classify messages, or summarize long pieces of content — all powered by your connected AI provider.

{% hint style="info" %}
[How to connect AI Providers?](/nected-docs/integrations/integrations-libraries/ai-providers.md)
{% endhint %}

<details>

<summary>How does it work?</summary>

When you add a Text Analysis Node inside your workflow, it takes your input text (or variables passed from earlier nodes), sends it to the configured AI model, and returns structured, machine-readable results such as sentiment scores, detected entities, or summarized text.

Each analysis type is pre-optimized for specific NLP tasks, so you don’t need to engineer prompts — just choose the analysis type, and Nected does the rest.

</details>

### **Configuration Tabs**

#### **1. Input Params**

This is where you define what the node should analyze and how.

**Connector dropdown:** Select your preferred AI connection. This determines which model and provider will power your text analysis.

**Assistant ID/Model name:** You can use a normal model or any of your AI assistants for this task. If you’re using a custom AI assistant from your provider (like OpenAI’s Assistant API), you can select its **Assistant ID**. This helps reuse a pre-configured assistant for specific tasks, instructions, or datasets. Or, you can use a normal AI model. In that case you only need to add the model name.

**Analysis Type:**\
Choose the kind of analysis you want the node to perform.\
Each type serves a different purpose:

| **Analysis Type**        | **What It Does**                                                                                           | **Common Use Cases**                                                            |
| ------------------------ | ---------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------- |
| **Sentiment Analysis**   | Identifies whether a given text expresses a positive, negative, or neutral tone.                           | Customer feedback evaluation, chat sentiment detection, social media monitoring |
| **Entity Recognition**   | Detects named entities such as people, companies, dates, and amounts, and classifies them into categories. | Extracting company names, policy numbers, product SKUs, or transaction details  |
| **Keywords Recognition** | Pulls out important keywords or phrases from the text.                                                     | Highlighting main topics, tagging content, indexing articles                    |
| **Text Classification**  | Categorizes the text into predefined labels or topics based on its content.                                | Ticket routing, content moderation, email sorting, intent detection             |
| **Summarization**        | Generates a concise summary of longer content while preserving meaning.                                    | Meeting notes condensation, news/article summaries, document overviews          |

**Text:** Enter the text you want to analyze. You can directly paste static text or pass dynamic values from previous workflow nodes using tokens.

For example:

```
{{.LoopBlock.output[1].customer_feedback}}
```

👉 [Learn more about using tokens in the editor →](https://docs.nected.ai/nected-docs/references/pre-configured-tokens/use-tokens-in-the-editor)

**Add Variable**

Variables allow you to pass additional instructions or identifiers to fine-tune analysis.

You can add:

* **Instructions** — to customize behavior (e.g., *“Only detect company and product entities.”*)
* **Thread ID** — to maintain context between multiple text analyses, especially in long-running conversations.

#### **2. Test Results**

Once you’ve set up the node, switch to the **Test Results** tab to view the AI’s response.\
When you click *Run Test*, the output appears here in **JSON format**, including:

* The **analysis result** (sentiment label, extracted entities, summary, etc.)
* **Processing time**
* **Reference ID**
* Optional **metadata**

You can toggle between *Raw*, *Pretty*, or *Table* views depending on how you want to inspect or debug the response.

For example:

```json
{
  "type": "sentiment",
  "label": "Positive",
  "confidence": 0.93
}
```

#### **3. Settings**

In this tab, you can adjust how the model executes the analysis.

* **Timeout for API (s):** Set how long to wait for an API response.
* **Timeout for Webhook/Cron (s):** Specify a limit for scheduled or webhook-based runs.
* **Continue on Error:** Allow the workflow to proceed even if the AI request fails.
* **Max Tokens:** Limit how long the response can be (use lower values for shorter summaries or classifications).
* **Temperature:** Control randomness — lower (0–0.3) for deterministic outputs, higher for creativity.
* **Max Retries:** Define how many times the node should retry in case of transient issues.
* **Metadata Required:** Enable this if you need extra information from the AI response (like token counts or processing stats).
* **Cache:** Save previous responses to speed up repeated analyses.
* **Time to Expire:** Set how long cached results remain valid.

### **What are the practical applications?**

Using the **Text Analysis Agent** in your workflow, you can automate the understanding and interpretation of text data at scale.

1. **Analyzing Customer Sentiment**\
   Detect positive or negative tones in feedback messages, and use a **Switch node** to route complaints to the support team automatically.
2. **Extracting Entities from Documents**\
   Run entity recognition to extract company names and monetary values from financial statements before storing them in your database.
3. **Summarizing Long Meeting Notes**\
   Feed entire transcripts into the Text Analysis Node with “Summarization” selected — the node returns a short, structured summary ready to be shared.
4. **Keyword Extraction for SEO**\
   Automatically identify the top keywords from new articles and save them in a database for indexing or analytics.