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

# How to configure AI Tagging

> Set up AI Tagging labels so the AI Agent summarizes, scores sentiment, and tags every conversation it handles in bitChat.

<Info>
  **What is AI Tagging?** When your AI Agent handles a conversation, AI Tagging reads it and writes a short summary, a sentiment, an intention, a few keyword tags, and one value for each label you define. Use the results to review tickets faster, build customer segments, and read the AI analytics charts.
</Info>

## How it works

AI Tagging runs only on conversations handled by the AI Agent. Conversations answered only by human agents are not tagged.

For each ticket the AI produces:

* **Summary** — a short English summary, shown in the ticket detail panel as the AI-generated ticket summary.
* **Sentiment** — `positive`, `neutral`, or `negative`.
* **Intention** — a short free-text description of what the customer wants. `order` is the special value for purchase intent.
* **AI tags** — 3–5 lowercase keywords describing the conversation.
* **Label values** — one value for each label you configure on the AI Tagging page (see below).

The results refresh automatically at several points during the conversation (after the customer's 2nd, 3rd, 5th, 8th, and 12th message, and so on) and once more when the ticket is resolved. Tagging is not real-time on every message, so a ticket that just started may not show tags yet.

<Note>
  AI Tagging does not consume AI credits. The number of labels you can create is capped per plan: **Free** 5, **Starter** 20, **Growth** 50, and **Pro Bundle** unlimited. See [pricing](https://bitbybit.studio/pricing/) for plan details.
</Note>

## Default labels

Every workspace starts with three labels. You can edit or delete any of them.

| Label | Values | What the AI does |
| :- | :- | :- |
| `language` | *none (free text)* | Writes the language of the conversation (for example `en`, `id`). |
| `priority` | `High`, `low`, `neutral` | Picks one, and that pick sets the ticket priority: `High` → **High**, `neutral` → **Medium**, `low` → **Low**. |
| `intention` | `Complaint`, `return`, `product`, `order`, `status` | Picks the one that best matches the conversation, or leaves it empty when none applies. |

<Note>
  The **Intention** chart in AI Agent analytics is built from the `intention` label. If you delete the label, that chart has less data.
</Note>

## Create a label

Create labels to track product interest, lead source, complaint type, or anything else you want to know about each conversation.

<Steps>
  <Step title="Add a new label">
    1. Navigate to **AI Studio** > **AI Tagging**.
    2. Click the **Add new tag** button in the top right corner.
  </Step>

  <Step title="Fill in the label">
    <AccordionGroup>
      <Accordion title="1. Label name">
        A short name, max 12 characters. This is the name you see in the ticket panel, filters, and analytics.

        * *Example:* `Laptop`
      </Accordion>

      <Accordion title="2. Label description">
        Tell the AI what the label means so it knows what to look for in the conversation. A specific description gives more consistent results.

        * *Example:* "Which laptop model the customer asks about"
      </Accordion>

      <Accordion title="3. Tag options (optional)">
        Turn this on and enter comma-separated options. The AI picks one of them for every ticket instead of creating its own values. If you leave it off, the AI writes a short free-text value instead.

        Tag options are not keywords the AI matches in the chat. The AI reads the whole conversation and chooses the option that fits best, even when the customer never types it word for word.

        * *Example:* `MacBook Air, Dell XPS, Lenovo ThinkPad`
      </Accordion>

      <Accordion title="4. Add to customer tags">
        Saves the label's value on the customer, so it applies to their future conversations too.
      </Accordion>

      <Accordion title="5. Allow multiple values">
        Enable for tags that gather multiple entries over time, such as customer interests.
      </Accordion>
    </AccordionGroup>

    <img src="https://mintcdn.com/asmaraku/HK3zE0cATE-negnJ/images/ai-studio/ai-tagging-label-form-updated.png?fit=max&auto=format&n=HK3zE0cATE-negnJ&q=85&s=af7a7a92f593f81ccebdf2e9184e1b27" alt="AI Tagging label form with Label name, Label description, the Tag options switch, and the Add to customer tags and Allow multiple values checkboxes" width="1320" height="762" data-path="images/ai-studio/ai-tagging-label-form-updated.png" />
  </Step>

  <Step title="Save">
    Click **Save**. The AI includes the label the next time it tags a conversation.
  </Step>
</Steps>

## Example labels

| Goal | Label name | Label description | Tag options |
| :- | :- | :- | :- |
| **Track products** | `Laptop` | Which laptop model the customer asks about | `MacBook Air, Dell XPS, Lenovo ThinkPad` |
| **Lead source** | `Lead source` | How the customer found the store | `Google, Facebook, Friend, Newsletter` |
| **Complaint type** | `Issue type` | Type of problem the customer reports | `Shipping, Defect, Payment, Login` |
| **Free-text value** | `Size asked` | Product size the customer asks about | *empty — the AI writes the size it sees* |

<Tip>
  Use tag options when you want clean groups for filters and segments (a fixed list is easier to count). Leave tag options off when the answer is open-ended, such as a product size or a city.
</Tip>

## Where you see the results

* **Ticket detail panel** in the bitChat inbox — the AI summary, sentiment, and AI tags for the open ticket.
  * The panel shows an **AI tagging** card. It shows the intention, sentiment, your label values, **Topics**, and **Key details** such as order number, tracking number, SKU, variant, and amount. Click a key detail to copy it, or click **Copy summary to ticket notes** to add the summary to the ticket notes.
* **Customers list** — the **Ticket tags** filter, and the **Ticket tags** condition in the segment builder. These work at the customer level and use the latest value of each label for that customer.
* **bitChat analytics overview** — the **Top 5 AI tags** card.
* **AI Tags analytics** page — AI tag trends across the period.
* **AI Agent analytics** — the **Intention** chart, built from the `intention` label.

The inbox chat list does not filter by AI tag, sentiment, or intention. To find customers by label value, use the **Ticket tags** filter in the Customers list.

## Related articles

* [AI Agent Analytics](/ai-studio/ai-agent-analytics)
* [Get started with AI Agent Studio](/ai-studio/getting-started-with-your-ai-agent)


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