AI-Powered Input Fields

Found in: Settings > AI Inputs

The AI-Powered Input Fields feature adds intelligent assistance to text fields throughout the Syncontrol platform. Users can select from pre-configured AI prompts that process field data along with optional file attachments, streamlining data entry and improving content quality.

Using AI Input Fields

Once an AI input is configured, it becomes available to users on the matching pages and fields.

Accessing the AI Assistant

When you navigate to a page that matches the configured URL pattern, an orange button with a sparkle icon appears next to input fields that match the CSS selector. Clicking this button reveals a dropdown with available AI prompts for that field.

AI Input button and dropdown

Each prompt shows its title (e.g., “AI Risk: Product description”). If you have administrator permissions, a pencil icon appears next to the prompt name, allowing you to quickly access the configuration interface.

Chat Interface

Selecting a prompt opens a chat-like popup where you can interact with the AI assistant. The system message indicates the assistant is ready, and you can type your instructions or questions in the text field at the bottom.

AI Chat interface

The AI processes your input along with any context data and attached files configured for that prompt. Responses appear in the chat, and you can continue the conversation to refine the output before inserting it into the field.

Prompt Preview (Administrators)

Administrators can access the Preview tab in the configuration interface to test how prompts will be processed. This tab shows:

Prompt Preview tab

Prompt Preview (with context substitution) Shows the prompt template with variables replaced by actual values from the current page context. This helps verify that your variable paths are correct and data is being pulled as expected.

Current Context Data (JSON) Displays the complete JSON object available for variable substitution. Use this to explore what data is accessible and find the correct paths for your prompt variables.

Matched Files Lists files that match the configured file filter, showing which documents will be available for AI processing.

Copy Context & Help / Copy Context Use these buttons to copy the context information to your clipboard. You can then paste this into ChatGPT or another AI assistant to help write and test prompt templates before saving them.

Working with AI Responses

Once you open the chat interface, you can interact with the AI in several ways:

Quick Generation Click the send button without typing anything to generate content based solely on the configured prompt and context data. The AI will process the prompt template with all available context and return a response.

Adding Instructions Type additional details or instructions before sending to guide the AI’s response. For example, you might specify a particular format, tone, or focus area.

AI response with Apply button

When the AI responds, you’ll see the generated content in a highlighted message. The Apply button appears at the bottom of the chat - clicking it inserts the AI’s response directly into the target field on the page.

Refining the Response If the initial response isn’t quite right, continue the conversation to modify it. Type follow-up instructions like “make it shorter”, “add more detail”, or even creative requests like “translate in pirate”.

Refined response in pirate speak

The AI maintains conversation context, so each refinement builds on the previous response. Continue iterating until you’re satisfied with the result, then click Apply to insert the final version into the field.

Key Capabilities

Intelligent Text Processing The assistant can enhance, correct, translate, or transform text based on the selected prompt configuration. This includes technical terminology correction, multi-language translation, and content enhancement.

File Attachment Support Documents and images can be attached to provide additional context. The system extracts text from PDFs, Word documents, and uses OCR (Optical Character Recognition) for images and scanned documents.

Context Awareness Prompts can access information about the current page, user, and entity being edited to provide more relevant suggestions.

Configuration

AI prompts are configured through the Settings > AI Inputs menu. The configuration interface has four tabs:

Basic Information

Field	Description
Title	Display name shown in the dropdown (e.g., “Field: Risk level conclusion”)
Description	Explanation of what this AI input does and when to use it
CSS Selector	CSS selector to target specific input fields or textareas on the page
URL Pattern	URL pattern where this AI input should be available

Understanding CSS Selectors

CSS selectors identify which input field on the page should display the AI assistant. The selector must match an input element, textarea, or editable field. In Syncontrol, most form fields have data attributes that make targeting straightforward.

To find the correct selector, use your browser’s developer tools (F12 or right-click > Inspect). Look for unique attributes on the input element you want to target. Syncontrol form fields typically use data-mobile-field-code attributes that correspond to the field’s configuration name.

Selector Type	Example	Matches
Attribute exact	`[data-mobile-field-code="risk_level_conclusion"]`	Field with exact code
Attribute contains	`[data-mobile-field-code*="risk"]`	Any field with “risk” in code
ID selector	`#description-field`	Element with specific ID
Class selector	`.inspection-notes`	Elements with specific class
Combined	`textarea[name="comments"]`	Textarea with specific name

Understanding URL Patterns

URL patterns control which pages display the AI input option. This prevents prompts from appearing where they’re not relevant. The pattern matches against the page path (everything after the domain name).

Use * as a wildcard to match any characters. Leave the field empty to make the AI input available on all pages. Multiple patterns can be separated by commas, and the AI input will appear if any pattern matches.

Pattern	Matches	Does Not Match
`/bookings/*`	`/bookings/123`, `/bookings/new`	`/booking/123` (no ‘s’)
`/accessable/, /jobs-to-review/`	Both `/accessable/456` and `/jobs-to-review/789`	`/other-page`
`/inspections/*/edit`	`/inspections/123/edit`	`/inspections/123/view`
(empty)	All pages	—

Prompt

The AI Prompt tab contains the instruction template that guides the AI’s response. Prompts can include:

Detailed instructions for the AI’s task
Examples of expected output
Context variables that get replaced with actual data

Variable Syntax

Two notation styles are supported for inserting dynamic data into prompts:

Style	Example	Description
Lodash Dot Notation	`{{context.model.id}}`	Model ID
	`{{context.model.user.name}}`	Nested property
	`{{files.document.content}}`	File content
JSONPath Syntax	`{{$.context.model.id}}`	Same as dot notation
	`{{$..name}}`	Find all “name” properties
	`{{$.files[0].data}}`	First file’s data
	`{{$.context.model[?(@.status=="active")]}}`	Filter by condition

Variables that don’t exist will be left as-is in the output. Objects and arrays are automatically JSON-formatted.

Lodash Dot Notation

Lodash notation uses simple dot-separated paths to access nested data. This is the easier syntax for straightforward data access. Start from the root object and chain property names with dots.

The available root objects depend on the context where the AI input is triggered. Common roots include context (page and model data), files (attached file contents), and field (current field value).

Example - Booking summary prompt:

Provide summary of the data for booking {{context.model.booking_uuid}}

Client Man Day rate: {{context.model.client.inspect_man_day_rate}}

JSONPath Syntax

JSONPath provides advanced querying capabilities when you need to filter, search, or access array elements. It always starts with $ representing the root object. Use this when Lodash notation isn’t sufficient for complex data structures.

JSONPath is particularly useful when working with arrays of files or when you need to find values across nested structures without knowing the exact path.

Common patterns:

Use Case	JSONPath	Description
First attached file	`{{$.files[0].data}}`	Access first file’s extracted text
All file names	`{{$..filename}}`	Find all filename properties recursively
Filter by status	`{{$.context.items[?(@.status=="approved")]}}`	Only items where status is “approved”
Specific array index	`{{$.context.defects[2].description}}`	Third defect’s description
Last item	`{{$.files[-1].content}}`	Last file’s content
Array length	`{{$.context.model.products.length}}`	Count of items in array
All items from array	`{{$.context.model.products[*].name}}`	All product names

Example - EU compliance report prompt:

# You are an EU product compliance consultant. You are working for your
# client who needs your help & expertise. You need to provide a structured
# report for a product that will be put on the EU market.

If possible, follow the approach and the format suggested in annex C of
the ISO 10377:2013 standard or in
https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32019D0417

--- DESCRIPTION ---
{{$.context.model.inspectionModel.data.product_description.value}}

--- APPLICABLE EU LEGISLATION ---
{{$.context.model.inspectionModel.data.applicable_eu_legislation}}

Example - Product listing prompt:

Provide summary of the data for booking {{context.model.booking_uuid}}

There are {{$.context.model.products.length}} products: {{$.context.model.products[*].product_name}}

Access Control

Configure which users can access this AI input using a permission matrix

Files

Configure file filters to control which files are available for AI processing. Filter groups use AND logic within a group.

Filter Criteria:

Criterion	Description
Filename	Match by filename (Equals, Contains, Starts with, Ends with)
File Size	Filter by file size
Folder	Match exact folder name
Subfolder	Match exact subfolder name
Properties (JSON Path)	Filter by file metadata properties
Object Table	Filter by associated object table

Multiple conditions can be combined with + AND to create complex filters. Use Add Filter Group to create OR logic between groups.

File Processing

The system supports various file formats for text extraction:

PDF Documents - Direct text extraction
Word Documents - DOC and DOCX parsing
Images - OCR via Azure Computer Vision
Scanned Documents - OCR processing for image-based PDFs

Use Cases

Inspection Forms

AI assistance helps with:

Risk level assessment and conclusion generation
Defect description enhancement
Technical terminology correction
Multi-language translation

Template Builder

The AI can suggest:

Field labels and descriptions
Validation rules
Help text and placeholders

Document Processing

Extract and process information from:

Booking documents
Technical specifications
Scanned forms

Usage Monitoring

The platform tracks AI usage for billing purposes:

Tokens consumed per request
API calls per user/tenant
OCR processing events