> For the complete documentation index, see [llms.txt](https://atlas.anywhere.now/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://atlas.anywhere.now/contact-center-configuration/building-your-callflow/via-dialogue-studio/ai-pathfinder-dc2025.01.md).

# Historical AI Pathfinder in DC2025.01

Historical AI Pathfinder setup for DC2025.01 only. Use the current AI-node guidance for later implementations.

{% hint style="warning" %}
Use this connection-string procedure only for the AI Pathfinder node in DC2025.01. It was replaced in DC2025.02a. It is not the current cache-based AI-node configuration path. Review legacy deployment compatibility before making changes, and use the current AI-node guidance for later implementations.
{% endhint %}

{% hint style="info" %}
Introduced in DC2025.01
{% endhint %}

Use **AI Pathfinder** in Dialogue Studio to route conversations by matching the customer’s question to the most relevant **Skill** description.

{% hint style="warning" %}
Important

This article applies to the AI Pathfinder node in DC2025.01. In DC2025.02a, AI Pathfinder was replaced by the AI nodes. For DC2025.02a guidance, see [AI Pathfinder](https://atlas.anywhere.now/contact-center-configuration/building-your-callflow/via-dialogue-studio/ai-nodes#ai-pathfinder).
{% endhint %}

## Prerequisites

* Administrator access to the **UCC Config** SharePoint site.
* An Azure OpenAI resource in your Azure subscription.\
  If you need to create one, follow [Microsoft’s guidance](https://learn.microsoft.com/azure/ai-services/openai/how-to/create-resource) and then return here.
* The Azure OpenAI **API key**, **endpoint**, and **model (deployment) name**—for example, `gpt‑4o‑mini`.

{% hint style="info" %}
Note

In DC2025.01 only non-reasoning models are supported, like gpt‑4o‑mini.
{% endhint %}

## Build your credentials JSON

The AI Pathfinder connection string is a single JSON object. Gather the three values from Azure AI Studio and build the string as shown below.

1. Open [Azure AI Studio](https://ai.azure.com/) and select your Azure OpenAI resource.
2. Copy the **Endpoint** (ends with `.openai.azure.com/`).
3. Create or copy an **API key**.
4. Under **Deployments**, copy the **name** you want to use—for example, `gpt‑4o‑mini`.
5. Replace the placeholders in the following snippet:

Credentials JSON

```
{"ApiKey":"YOUR_API_KEY","Endpoint":"https://YOUR_RESOURCE.openai.azure.com/","Deployment":"YOUR_MODEL_NAME"}
```

{% hint style="warning" %}
Common errors

* Use the **Endpoint** URL, not the **TargetUri** shown in some REST examples.
* Enter the **(deployment) name**, such as `gpt‑4o‑mini`. Do *not* use the friendly name of your Azure deployment object or the Model name.
  {% endhint %}

## Store credentials and system prompt

Add the credentials JSON and the system prompt to the **Plugin Settings** list so that AI Pathfinder can call Azure OpenAI.

1. Select **UCC Config › Plugin Settings**.
2. Enter the rows shown in the following table.

| Setting                                                                                                 | Value                                                                                                                                         | Description                                                                                       |
| ------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------- |
| <p><strong>Name:</strong> Credentials<br><strong>Scope:</strong> DialogueStudioAiNodes</p>              | `{"ApiKey":"<API_KEY>","Endpoint":"<ENDPOINT>","Deployment":"<MODEL_NAME>"}`                                                                  | Connection string AI Pathfinder uses to call Azure OpenAI.                                        |
| <p><strong>Name:</strong> AiPathfinderSystemPrompt<br><strong>Scope:</strong> DialogueStudioAiNodes</p> | <p>Prompt<br><a href="#configure-historical-ai-pathfinder-in-dc2025.01-code-3">Configure Historical AI Pathfinder in DC2025.01 Code 3</a></p> | System prompt sent to Azure OpenAI. AI Pathfinder replaces `{0}` with enabled skill descriptions. |

#### Configure Historical AI Pathfinder in DC2025.01 Code 3

Source row: **Name: AiPathfinderSystemPromptScope: DialogueStudioAiNodes**.

```
Match the customer transcript to one of the provided departments in the JSON object below. If no match is found, return "Other".

Consider each department and evaluate how well the input transcript corresponds to it based on its context, keywords, or implied needs. If no clear match can be made, categorize as "Other" and route to the following skill: Other.

# Steps
1. **Analyze the transcript**: Inspect the details provided in the transcript for clues, such as keywords, phrases, or implied topics that correspond to specific departments.
2. **Compare to departments**: Match the context of the transcript to available departments in the department list.

# Output Format
Only respond in a string.

# Notes
- Ensure the string response is valid and exactly matches the Title from the department list.
- Consider edge cases, like vague or unclear transcripts, and ensure the motivation sufficiently supports the decision to choose "Other" and route to the following skill: Other.

# Departments
{0}
```

## Enable skills for AI Pathfinder

AI Pathfinder only considers skills that you explicitly enable.

1. Select **UCC Config › Skills**.
2. For each skill, enter a clear **Description** that explains when the skill should be used.
3. Select the **PathFinderEnabled** checkbox for every skill that AI Pathfinder may choose.

### Examples of clear skill descriptions

Use concise, customer-focused language so the model can quickly identify the right department. Avoid internal jargon or overlapping scopes.

| Skill title        | Example description                                                                              |
| ------------------ | ------------------------------------------------------------------------------------------------ |
| Order Status       | Requests for tracking numbers, delivery dates, shipping delays, or changes to an existing order. |
| Billing & Payments | Questions about invoices, payment methods, refunds, duplicate charges, or tax documentation.     |
| Technical Support  | Issues installing, updating, or using the product; error messages; performance problems.         |
| Sales & Upgrades   | Inquiries about new licences, feature packages, pricing, or product comparisons.                 |
| Account Management | Requests to update contact details, change passwords, or review security settings.               |

{% hint style="success" %}
Tip

Start each description with the type of request, then list common examples. This structure helps Azure OpenAI find strong keyword matches.
{% endhint %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://atlas.anywhere.now/contact-center-configuration/building-your-callflow/via-dialogue-studio/ai-pathfinder-dc2025.01.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.