Real-Life Demo: AI-Assisted D365 F&O Development

Watch how an AI agent uses the Artifact MCP Server to complete a D365 F&O customization task

📋 The Task

User Request:

"Create a new field 'External name' on vendor group table if it is not exist, and add it to form general tab"

What You'll See:

  • ✅ AI using search_fo_artifacts MCP tool to find standard D365 artifacts
  • ✅ AI reading standard XML files via fullLocalPath
  • ✅ AI searching user's workspace for existing customizations
  • ✅ AI generating complete table and form extensions following D365 patterns
  • ✅ Complete implementation delivered in a single session
Step 1

Initial Search for VendGroup Table

Step 1: Initial MCP tool call to search for VendGroup table

What's Happening:

  • AI breaks down the user request into actionable steps
  • First action: Find the VendGroup table in standard D365
  • Uses search_fo_artifacts with query: "VendGroup vendor group table"

Result:

Found 5 artifacts, including VendVendorGroupStaging table. AI recognizes this is a staging table, not the actual VendGroup table.

🔍 Key Observation: The AI demonstrates intelligent refinement - it recognizes the initial results aren't quite right and prepares to search again.
Step 2

Refined Search with Better Query

Step 2: Refined search for actual VendGroup table

What's Happening:

  • AI recognizes staging table isn't what's needed
  • Refines query to: "VendGroup table vendor classification group"
  • Uses more specific terminology to find the base table

Result:

Found the actual VendGroup table with fullLocalPath to the XML file.

💡 MCP Tool Value: The tool provides the exact file path to standard D365 metadata that the AI couldn't otherwise access.
Step 3

Search for VendGroup Form

Step 3: Search for VendGroup form

What's Happening:

  • AI now needs to find the VendGroup form to add the field to the General tab
  • Uses search_fo_artifacts with query: "VendGroup form vendor classification"
  • Plans next steps: check for existing extensions, read standard implementations

Result:

Found VendGroup Form artifact with full metadata and file path.

🎯 Workflow Pattern: AI is following a systematic approach - find table, find form, check customizations, then implement.
Step 4

Reading Standard Files & Checking Workspace

Step 4: Reading F&O asset files via fullLocalPath

What's Happening:

  • Critical moment: AI uses the fullLocalPath provided by MCP tool
  • Reads actual VendGroup.xml files to understand structure
  • Searches user's workspace for existing VendGroup extensions
  • No existing extension found - will need to create new files
🔑 This is the Core Value: The MCP tool bridges the gap between AI and local D365 metadata. The AI can now read the actual standard implementation instead of guessing or hallucinating.
Step 5

Understanding Form Structure & Patterns

Step 5: AI researching form structure and patterns

What's Happening:

  • AI reads CustParameters table extension to understand extension patterns
  • Searches for "General tab" structure in VendGroup form
  • Uses grep to find General tab naming conventions
  • Reads CustTable form to understand similar implementations

Result:

AI learns D365 patterns: TabGeneral naming, FastTab structure, field group organization.

🧠 Context-Aware Development: The AI isn't just writing code - it's learning from existing patterns in both standard D365 and user's customizations.
Step 6

Creating Extensions

Step 6: AI creates table and form extensions

What's Happening:

  • AI creates VendGroup table extension with new field
  • AI creates VendGroup form extension adding field to General tab
  • Both files shown as "new" with line counts (+25 lines each)
  • Proper XML structure following D365 conventions

Key Details:

  • Field uses Name extended data type (EDT)
  • Form control bound to VendGroup data source
  • Placed in TabGeneral (General tab) as requested
Step 7

Complete Summary & Next Steps

Step 7: Final summary and next steps

What's Delivered:

1. Table Extension
  • Added field to VendGroup table
  • Field Type: String (using Name EDT)
  • Label: "External name"
  • Help Text: "External name for vendor group"
2. Form Extension
  • Added field to VendGroup form
  • Placed in TabGeneral (General tab)
  • Bound to VendGroup data source
3. Next Steps Provided
  1. Build the model in Visual Studio
  2. Synchronize database to create the new field
  3. Test in UI (Accounts Payable > Setup > Vendor groups)
  4. Add to source control when verified
✅ Task Complete: From a simple natural language request to complete, correct implementation - all in a single AI session with MCP tool assistance.

🎓 What This Demonstrates

🔍 Semantic Search Works

The AI successfully found relevant D365 artifacts using natural language queries, even refining searches when needed.

📁 File Access is Critical

The fullLocalPath field enabled the AI to read actual standard implementations instead of guessing structure.

🧩 Pattern Recognition

AI learned from both standard D365 code and user's existing customizations to generate context-aware solutions.

⚡ Iterative Workflow

The tool supported an exploratory workflow - search, read, search again, read more, implement.

✨ Complete Implementation

From request to working code in minutes, with proper XML structure, naming conventions, and actionable next steps.

🎯 No Hallucination

AI didn't make up field names, table structures, or form designs - it looked them up using real metadata.

Try It Yourself

Ready to experience AI-assisted D365 F&O development?

Learn About MCP Server View on GitHub