Usage Tips & Best Practices

This document summarises prompting techniques for the AI Assistants, cross‑assistant collaboration workflows and team usage guidelines. It helps you use the AI Page Assistant, AI Variable Assistant, AI Table Assistant, AI Model Assistant, AI Image Assistant, AI Script Assistant and AI Smart Binding more efficiently.

1. General Prompting Techniques

1. Use the "5W1H" Framework

A clear requirement description should include:

• What: What function needs to be implemented?
• Why: Business background and purpose.
• Who: Which users are targeted?
• When: Trigger timing, execution time or refresh frequency.
• Where: Application scenario, module entry point or device scope.
• How: Specific implementation method, rules or constraints.

Not recommended:

Create a monitoring page

Recommended:

What: Create an injection moulding machine monitoring page.
Why: For real‑time equipment status monitoring on the shop floor.
Who: Workshop operators and shift leaders.
When: 7x24 real‑time display, data refreshes every 2 seconds.
Where: Workshop large screen, resolution 1920x1080.
How: Include temperature curves, pressure gauges, production count, status indicator lights and an alarm list.

2. Describe Complex Requirements in Layers

For complex requirements, use a structure of "overall goal -> functional modules -> detailed rules".

Layer 1: Overall goal

Create an MES workstation page for work order management and data collection on the production floor.

Layer 2: Functional modules

The page is divided into three main areas:
1. Work order information area: displays current work order details.
2. Data collection area: barcode scanning, input, confirmation.
3. History area: shows the last 10 operation records.

Layer 3: Detailed rules

The work order information area contains:
- Work order number, displayed in a large font.
- Product model, target quantity, completed quantity.
- Work order status, colour‑coded: green for in progress, yellow for paused, red for abnormal.

3. Use Specific Values and Units

Not recommended:

Show some temperature data

Recommended:

Show data from 4 temperature sensors, range 0–200 degrees Celsius, with one decimal place. Display in red when above 180 degrees.

4. Clearly Specify Input Materials and References

The AI Page Assistant, AI Variable Assistant, AI Table Assistant, AI Model Assistant and AI Image Assistant all benefit from more explicit input materials to produce more stable results.

Using the uploaded dashboard screenshot as a reference, adopt its layout and dark tech style,
but change the content to OEE, production trends, equipment status and alarm information for production line 2.

Using the uploaded CSV point table, which contains variable names, register addresses, data types and read/write permissions.
Please parse it according to the Modbus TCP protocol. Device IP is 192.168.1.100, port is 502.

5. Specify Priorities

For multiple requirements, it is recommended to label priorities to help the AI focus on the core functionality first.

Must include:
- Real‑time temperature display
- Equipment status indication
- Alarm list

Important but can be adjusted later:
- Temperature history curve
- Shift production statistics

Optional:
- Placeholder for equipment 3D model

2. Cross‑assistant recommendation workflow

1. From Device Integration to Page Display

Suitable for new device integration, production line monitoring pages and project delivery scenarios.

1. Use the AI Variable Assistant to generate the device point table and form standard variables.
2. Use the AI Table Assistant to create history databases, aggregate statistical tables or business data tables.
3. Use the AI Page Assistant to generate monitoring pages, dashboards or large screens and leave placeholders where equipment display is required.
4. Use the AI Model Assistant to generate 3D equipment models, and the AI Image Assistant to generate or enhance equipment image assets.
5. Use the AI Smart Binding to bind page components to variables and data tables.
6. Use the AI Script Assistant to add business logic, system integration and automation control.
7. Validate data, interactions, scripts and page presentation in a test environment.

2. Prototype‑First Workflow

Suitable for bid demonstrations, customer communication and solution reviews.

1. Use the AI Page Assistant in Quick Mode to generate a page prototype.
2. Use the AI Model Assistant to generate 3D models of key equipment and reference them on the page.
3. Use the AI Page Assistant for local regeneration based on customer feedback.
4. After the requirements stabilise, supplement variables, tables, bindings and script logic.

3. Delivery‑First Workflow

Suitable for projects with strict standards that require real data and maintainability.

1. First organise variable naming conventions, point tables and data archiving rules.
2. Use the AI Variable Assistant to complete standardised variable parsing and generation.
3. Use the AI Table Assistant to configure history databases, operational and business data tables.
4. Use the AI Page Assistant in Guided Mode to generate pages.
5. Use AI Smart Binding and review the binding suggestions item by item.
6. After generating scripts with the AI Script Assistant, perform testing and validation.

3. Best Practices by Assistant Type

1. AI Page Assistant

Choose the Right Generation Mode

• Quick Mode: Suitable for early‑stage concept design, proof of concept and page prototyping.
• Guided Mode: Suitable for delivery‑oriented projects with clear requirements and strict standards.

Page Generation Formula

Page theme + Usage scenario + Visual style + Layout structure + Data to display + Interaction requirements

Example: Equipment monitoring page

Create an injection moulding machine monitoring page for real‑time equipment status display on a workshop large screen.
Use a dark tech style, resolution 1920x1080.

Layout:
- Top: equipment name, current shift, running status and time.
- Left: temperature curves for the three barrel zones.
- Middle: injection pressure gauge, clamping force gauge and equipment schematic.
- Right: production count, cycle time, yield rate and alarm list.
- Bottom: start, stop, reset, emergency stop buttons.

Requirements:
- Status colours: green for running, yellow for standby, red for fault.
- Use large fonts for key data.
- Show a maximum of 10 alarms.

Use Reference Images and Asset Templates

If you have a target style, upload a reference image or select an asset template first, then explain which parts should be referenced.

Please refer to the three‑column layout and dark colour scheme of the image I uploaded,
but do not copy the text content directly.
Change the page theme to packaging line monitoring.
Display the production line flow in the middle area, and show alarms and work order information on the right.

Make Good Use of Local Regeneration

If the overall page is satisfactory but a local area is not, you do not need to regenerate the whole page. First select the target area, then describe the modifications.

Adjust only the area I have selected:
1. Change the alarm list to a card list.
2. Each alarm card should show time, device name, alarm level and handling status.
3. Emphasise high‑level alarms in red.

2. AI Variable Assistant

Point Table Preparation Suggestions

The AI can parse point tables that are not completely standard, but standardised headers significantly improve accuracy. It is recommended that point tables include at least:

• Variable name
• Address or register
• Data type
• Read/write permission
• Unit
• Description

Provide Protocol Parameters in the First Conversation

If you already know the device communication parameters, state them together in the first conversation.

Please import this CSV point table.
Protocol: Modbus TCP, device IP: 192.168.1.100, port: 502, station number: 1.
Add the prefix Line2_ to all variable names and keep the original descriptions.
Please generate an import plan first, and execute after confirmation.

Small‑Batch Variable Creation

Create an S7 device channel with IP 10.0.0.5, Rack 0, Slot 1.
Variables include:
- Temperature, DB1.DBD0, Float, read‑only
- Pressure, DB1.DBD4, Float, read‑only
- Start/stop status, DB1.DBX8.0, Bool, read‑only

Variable Naming Convention

It is recommended that teams adopt a unified variable naming scheme to facilitate subsequent smart binding and table creation.

Device Type_Device ID_Parameter Type_Parameter Number

Examples:
- Injection_M1_Temperature_01
- Injection_M1_Pressure_Main
- Packaging_L1_Speed_Conveyor

3. AI Table Assistant

First Determine the Table Type

• History Database: For continuous acquisition, trend curves, tracing of equipment operation data.
• Data Management Tables: For business archiving, statistical analysis, master data maintenance and aggregation calculations.

History Database Archiving Prompt

Archive all temperature and pressure variables on production line 2 to the history database.
Use change recording for temperature variables, deadband 0.5 degrees Celsius.
Use timed recording every 10 seconds for pressure variables.
Automatically group by variable type, generate a history group plan, and let me confirm before execution.

Data Management Table Creation Prompt

Create a shift energy consumption statistics table for production line 2.
Fields include:
- Shift name, String
- Start time, DateTime
- End time, DateTime
- Electricity consumption, Float, unit kWh
- Water consumption, Float, unit t
- Compressed air consumption, Float, unit Nm3

Match data sources from the electricity meters, water meters and air consumption variables on line 2.
Generate statistical rules for morning, afternoon and night shifts.
Please generate the plan first.

Organize an Attachment When There Are More Than Ten Fields

If there are many fields, first organise them into a CSV and then upload it to the AI. The field list should include field name, data type, whether required, default value, source variable and remarks.

4. AI Model Assistant

Text‑to‑3D Prompt Formula

Industry scenario + Equipment type + Structural features + Material/color + Style requirements

Example:

A vertical storage tank for a chemical plant, made of stainless steel, smooth surface without welds. It is equipped with a safety valve and a level gauge on top and has a conical bottom. The tank features a silver-gray metallic luster with a realistic industrial style. The pipelines and instruments can be optionally added in the background to convey a sense of engineering rigor.

Image‑to‑3D Image Preparation Suggestions

• Keep the background simple, preferably a solid colour.
• Only one subject device in the image.
• The subject should not be too small; the outline should be clear.
• Avoid text, watermarks and obstructions as much as possible.
• Use clear images – the clearer the image, the better the result, resolution no lower than 128x128.

Polygon Count Selection Suggestions

• Quick preview: start with a lower polygon count.
• Key page display: choose medium or high polygon count.
• Multiple devices on the same screen: control the polygon count to avoid affecting page performance.

5. AI Image Assistant

Equipment Image Prompt Formula

Equipment type + Structural features + Display angle + Material/colour + Style requirements + Page usage

Example: Single equipment image

Generate an injection moulding machine image,
two-thirds profile view,
blue‑grey industrial style,
complete equipment structure, clean background,
suitable for display on a configuration page.

Example: Metallic equipment display image

Generate a vertical storage tank image,
front view,
stainless steel material,
realistic display style,
clean background,
suitable for the equipment display area on a configuration page.

Reference Image Enhancement and Background Removal

If you already have an on‑site equipment photo, upload the reference image first, then describe what to preserve, what to improve and the final purpose.

Using this equipment image as a reference,
preserve the overall outline and front view,
enhance the equipment texture,
remove the cluttered background,
make the background clean,
for use in the equipment display area of a configuration page.

Keep Equipment Images Consistent with Page Style

Before generating an image, decide on the page style, then clearly specify the equipment angle, material, colour, background complexity and final purpose. Equipment images used on configuration pages should highlight the equipment subject and avoid excessive decoration that might interfere with data readability.

6. AI Script Assistant

Modular Script Development

Break complex scripts into multiple modules, generate them separately and then combine.

Data acquisition module

Write a data acquisition class DataFetcher:
- Method: GetFromAPI(url, params)
- Function: call an HTTP API to get data
- Return: JSON object
- Exception handling: retry 3 times on timeout, return null on failure

Data processing module

Write a data processing class DataProcessor:
- Method: ParseJSON(jsonString)
- Function: parse JSON and extract specified fields
- Return: Dictionary<string, object>
- Exception handling: log parsing failures

Data storage module

Write a data storage class DataStorage:
- Method: SaveToDatabase(data)
- Function: batch insert into the database
- Return: number of successful records
- Exception handling: transaction rollback

Specify Trigger Conditions and Exception Handling

When the variable Work Order ID changes, call the MES interface to obtain work order information.
Requirements:
- If WorkOrderID is empty, do not call the interface.
- Interface timeout: 10 seconds.
- On failure, log the error without affecting the main flow.
- After success, write to variables ProductModel and TargetQuantity.

Performance Optimization Prompt

Please optimise the performance of this script:
[paste original code]

Requirements:
- Avoid writing to the database one record at a time inside a loop.
- Use batch processing.
- Process 100 records per batch.
- Add exception handling and logging.

7. AI Smart Binding

Precisely Select the Variable Scope

When selecting the variable scope, try to choose the smallest possible scope. If the page involves only a single device, do not select the entire workshop.

Page Type	Recommended Channel Scope	Matching Effect
Single device monitoring page	Select that device channel	Most stable
Multiple devices of the same type	Select the device group channel	Relatively stable
Workshop comprehensive dashboard	Select the workshop channel	Needs focused review
Plant‑wide large screen	Select multiple workshop channels	Needs item‑by‑item confirmation

Review Everything the First Time You Use It

AI Smart Binding provides suggestions. Before applying, check the component meaning, recommended variable and confidence level. Items with low confidence, warnings or failures require manual confirmation.

Batch Process Similar Pages

1. Use Smart Binding on the first page and review carefully.
2. Save as a page template or copy the page.
3. Modify the device ID and channel scope.
4. Rerun Smart Binding.
5. Spot‑check key data points.

4. Prompt Template Library

Injection Molding Machine Monitoring Page

Create an injection moulding machine monitoring page, industrial dark style.

Equipment information:
- Equipment name: [device ID]
- Monitoring parameters: 3 barrel temperatures, mould temperature, injection pressure, clamping force
- Production data: cycle time, production count, yield rate
- Control buttons: start, stop, emergency stop, reset

Layout requirements:
- Top: equipment name, running status, current time
- Left: temperature curves
- Middle: pressure and clamping force gauges
- Right: production data and control buttons

Packaging Line Monitoring Page

Create a packaging line monitoring page.
Line composition: feeder -> packaging machine -> sealer -> labeller -> case packer.

Display content:
- Line flow diagram, horizontal layout, showing 5 devices.
- For each device, show device name, running status, current speed and production count.
- At the bottom, show total line production count, yield rate and current work order information.

Interactive functions:
- Click on a device icon to pop up detailed parameters.
- Real‑time refresh, refresh interval 2 seconds.

Energy Monitoring Large Screen

Create an energy management large screen, resolution 1920x1080.
Monitoring scope: 5 workshops, each monitoring electricity, water and compressed air.

Layout:
- Top: total energy, today's energy, this month's energy, this year's energy.
- Left: last 24‑hour energy trend and workshop energy comparison.
- Middle: factory floor plan and real‑time power curve.
- Right: energy type percentage, energy ranking and alarm information.

Theme:
- Green environmental style.
- Dark background.
- Highlight colours: cyan/blue and green.

Variable Point Table Import

Please parse and import the uploaded CSV point table.
Protocol: Modbus TCP, device IP: 192.168.1.100, port: 502, station number: 1.
Identify variable names, register addresses, data types, read/write permissions and descriptions.
If any necessary parameters are missing, first ask me to supplement them via a card.
Before writing, please show me the import plan.

History Database Archiving

Please create a history database archiving plan for production line 2.
Archiving objects: all temperature, pressure and speed variables.
Use change recording for temperature, deadband 0.5.
Use timed recording for pressure and speed, interval 10 seconds.
Automatically group by variable type, generate a history group plan and let me confirm.

3D Equipment Model

Generate a 3D equipment model for display on a configuration page:
Chemical industry, vertical storage tank, stainless steel material,
pipe connection on top, support base at the bottom,
clear structure, realistic industrial style.

Equipment Image Asset

Generate an air compressor image for display on a configuration page.
Front‑side combined angle,
blue‑grey industrial style,
complete structure, clean background,
suitable for the main display area of an equipment monitoring page.

5. Solutions to Common Problems

Problem 1: AI‑generated page does not meet expectations

Common causes:

• Requirement description is not clear enough.
• Layout, resolution or style requirements are missing.
• Core data and priorities are not stated.

Solutions:

1. Redescribe requirements using the "5W1H" framework.
2. Upload a reference image or select an asset template.
3. Use Guided Mode to complete the requirements.
4. For local problems, use local regeneration on the selected area.

Problem 2: Variable assistant parsing is inaccurate

Common causes:

• Point table headers are not standardized.
• Protocol, IP, port and other channel parameters are missing.
• Variable names, addresses or data types are mixed into remarks.

Solutions:

1. Organize clear headers as much as possible.
2. Provide protocol and device parameters in the first conversation.
3. Let the AI output an import plan first, then confirm execution.
4. Directly ask the AI to correct any parsing errors in the conversation.

Problem 3: The table creation scheme is inconsistent with the business definition

Common causes:

• Variables have not been created yet.
• Requirement description is unclear.
• Variable scope is too broad or variable names are unclear.

Solutions:

1. First create the variables in the variable management module.
2. Clearly specify the table creation requirements.
3. Review the plan card before execution.

Problem 4: Low accuracy of Smart Binding

Common causes:

• Variable names are not standardized.
• The selected channel scope is too broad.
• Variables lack descriptive information.

Solutions:

1. Improve variable naming and use meaningful names.
2. Add detailed descriptions to variables.
3. Precisely select the device channel scope.
4. Use variable grouping.
5. Review low‑confidence results item by item.

Problem 5: Model or image assets are unsuitable for the page

Common causes:

• The prompt only describes the subject, not the page usage.
• Style, color and composition are inconsistent with the page.
• The equipment image subject is incomplete, the background is cluttered, or the texture is unsuitable for a configuration page.
• The image or model has too many details, affecting page performance or data readability.

Solutions:

1. Clarify the asset usage, e.g. main equipment image, schematic, image component display or 3D component display.
2. Specify the page style, equipment angle, material color and background requirements.
3. First validate the model with a lower polygon count, then increase detail.
4. Prioritize the equipment subject in images; for on‑site photos, use reference image enhancement and background removal.
5. After generation, place the asset on a real page to check for stretching, cropping, resolution and style consistency.

Problem 6: Script has performance or stability issues

Common causes:

• Time‑consuming operations inside a loop.
• Batch processing is not used.
• Missing exception handling and logging.
• Timeout, retry and failure strategies not specified.

Solutions:

1. Ask the AI: "How can I optimize the performance of this code?"
2. Specify performance requirements, e.g. processing 1000 data records must complete within 1 second.
3. Indicate optimization directions, e.g. batch processing, caching, asynchronous calls.
4. Validate the script in a test environment.

Problem 7: Credits are consumed too quickly

Common causes:

• Frequent repeated generation and conversation edits.
• Unclear requirement descriptions leading to many attempts.
• Not reusing existing results.
• Each conversation may consume credits.

Solutions:

1. Carefully check the requirement description before the first generation – state everything clearly in one go.
2. For complex tasks, let the AI output a plan first, then execute the generation or write.
3. Save excellent pages, prompts, variable naming conventions and table creation templates.
4. For local page problems, use local regeneration instead of regenerating the whole page repeatedly.
5. Reuse similar pages, scripts, equipment models and equipment image assets.

6. Team Collaboration Suggestions

1. Build a Team Prompt Library

Accumulate commonly used prompts by industry, page type, script type, variable protocol, table type, equipment model type and equipment image type to reduce repeated trial and error.

2. Establish Naming and Field Conventions

It is recommended to unify the following conventions:

• Variable naming convention
• Channel naming convention
• Page naming convention
• Data table naming convention
• Field naming and data type convention
• Asset naming and categorization convention

3. Establish a Review Mechanism

AI‑generated results should be manually reviewed. Content that affects production data or operational logic – especially variable generation, table creation, scripts, smart Binding – should be validated at least once in a test environment.

4. Accumulate Reusable Assets

Organize validated pages, scripts, data table structures, history database strategies, equipment models, equipment image assets and prompts into team assets. Subsequent projects can reuse them quickly.