# Custom LLM Configuration

Platform ID: FU-10146\
Document Version: 1.0\
Date: 13-02-2026<br>

### 1. Introduction

Custom LLM Configuration enables organizations to configure how artificial intelligence operates within structured workflows inside Unifize.

This feature allows administrators to define organization-wide AI behavior, configure workflow-level instructions, and select AI models for controlled AI-assisted output generation.

AI operates only within defined workflows and only when explicitly triggered by a user.

Custom LLM Configuration allows organizations to:

* Define an organization-level System Prompt
* Configure workflow-level Add Prompts
* Select approved AI models
* Control how AI suggestions are generated
* Maintain structured and governed AI output\ <br>

This ensures AI-generated summaries, analyses, and suggestions are:

* Consistent with organizational expectations
* Scoped to workflow data
* Controlled through explicit configuration\ <br>

***

### 2. Capabilities

With Custom LLM Configuration, users can:

* Configure an organization-wide System Prompt
* Add contextual prompts within specific workflow fields
* Select from approved AI models
* Upload supported files as AI context (PDF, .md, .json)
* Generate structured summaries and analyses
* Ensure prompt hierarchy is consistently enforced\ <br>

Administrators manage configuration.\
\
Organization members use AI within permitted workflows.

***

### 3. User Journey

### The configuration and usage flow involves both Admins and Organization Members.

#### Step 1: Access Organization Settings (Admin Only)

Admin logs into Unifize.

Navigate to:

Profile → Org Settings → Org Details → System Prompt

Only Admin users can edit this configuration.

If a non-admin attempts to modify the System Prompt, the action is restricted.

<br>

#### Step 2: Configure System Prompt

The Admin can:

* Enter organization-level instructions for AI behavior
* Use structured formatting or multilingual content
* Save the System Prompt\ <br>

Once saved:

* A confirmation message appears
* Configuration persists across sessions
* All admins see the updated prompt\ <br>

The System Prompt applies across the organization and governs AI behavior globally.

<br>

#### Step 3: Configure Workflow-Level Settings

Within a workflow that contains an AI-enabled field:

1. Navigate to the field settings
2. Add an Add Prompt
3. Select an AI Model
4. Save configuration\ <br>

The Add Prompt:

* Provides field-specific instructions
* Works in combination with the System Prompt\ <br>

Model selection:

* Applies only to the selected field
* Does not affect other fields\ <br>

<br>

#### Step 4: Upload Context Files (Optional)

Within field configuration, users may upload:

* PDF
* .md
* .json\ <br>

Supported files are included in AI execution context.

Unsupported formats are rejected with a validation error.

<br>

#### Step 5: Invoke AI

Within a record containing an AI-enabled field:

1. User enters relevant input
2. Clicks the AI action button
3. AI generates structured output\ <br>

AI execution uses:

* Record data
* System Prompt (if configured)
* Add Prompt (if configured)
* Selected AI model\ <br>

AI runs only when explicitly triggered by the user.

<br>

#### Step 6: View Generated Output

Generated output:

* Appears within the workflow field
* Follows structured system format
* Reflects configured prompts
* May vary slightly in wording between executions\ <br>

If the System Prompt is updated, future executions reflect the updated instructions.

If no System Prompt exists, AI continues to function using field-level configuration.

<br>

### 4. Prompt Governance

1. AI behavior follows a defined hierarchy:
2. System Prompt (Organization Level)
3. Add Prompt (Workflow Level)
4. Record Data / User Input
5. Model-Specific Behavior
6. The System Prompt governs tone, structure, and overall behavior.
7. Changing the AI model affects output characteristics but does not override prompt governance.
8. Customers are responsible for validating AI use within their processes.
9. These limitations are inherent to AI-assisted systems.

<br>


---

# Agent Instructions: Querying This Documentation

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

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.unifize.com/user-guide/custom-llm-configuration.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
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.