Drupal AI Module Migration Strategy

Executive Summary

Comprehensive analysis of _DrupalSource modules reveals 35-45% code reduction opportunity by migrating custom AI implementations to the standardized Drupal AI module framework. This migration will improve maintainability, leverage community contributions, and align with Drupal best practices.

Analysis Results

Total Impact Assessment

12 modules analyzed with custom AI functionality
Average code reduction: 40% across all modules
Most impactful migrations: Provider modules (55-70% reduction)
Timeline: 4-phase implementation over 8-12 weeks

Module-by-Module Breakdown

High Impact Migrations (55-70% reduction)

ai_provider_langchain - 70% reduction potential

Current: 492 lines of custom HTTP client, API management, MLX optimization
Drupal AI equivalent: Provider plugin with automatic HTTP handling
Key replacements:
- Custom HTTP client → AI module's standardized HTTP service
- API key management → Built-in key integration
- Custom MLX detection → Provider capability system
Migration effort: 2-3 days

ai_provider_apple - 65% reduction potential

Current: Custom provider implementation with Apple-specific optimizations
Drupal AI equivalent: Provider plugin with Apple Intelligence integration
Key replacements:
- Custom provider registration → Plugin discovery system
- Manual API integration → Standardized provider interface
Migration effort: 2-3 days

Medium Impact Migrations (35-50% reduction)

llm (Core Module) - 45% reduction potential

Current: 701 lines in services.yml with extensive custom provider management
Drupal AI equivalent: Leverage AI module's provider system
Key replacements:
- Custom provider manager → @plugin.manager.ai_provider
- Manual provider registration → Automatic plugin discovery
- Custom chat interfaces → Standardized AI module interfaces
Business logic preserved: Conversation management, usage tracking, cost calculation
Migration effort: 5-7 days (largest module)

ai_agent_crewai - 45% reduction potential

Current: 190 lines with simplified orchestration service
Drupal AI equivalent: AI automators and workflow integration
Key replacements:
- Custom agent roles → AI automator plugins
- Manual task processing → Workflow-driven execution
Migration effort: 3-4 days

mcp_registry - 40% reduction potential

Current: Custom MCP server management and discovery
Drupal AI equivalent: Provider plugin system for MCP servers
Key replacements:
- Custom server registration → Plugin discovery
- Manual health monitoring → Built-in provider health checks
Migration effort: 3-4 days

Lower Impact Migrations (20-35% reduction)

gov_compliance - 35% reduction potential

Current: Policy-driven security framework
Drupal AI equivalent: AI logging and audit integration
Preserved: Custom compliance policies and government-specific features

ai_agent_huggingface - 30% reduction potential

Current: Model discovery and pipeline management
Drupal AI equivalent: Provider system for Hugging Face models

Alternative modules - 25% average reduction

api_normalizer: OpenAPI import → Entity API integration
alternative_services: Service discovery → AI provider discovery
ai_agent_orchestra: ECA integration → AI automators
recipe_onboarding: Tour integration → AI-powered onboarding

Migration Strategy

Phase 1: Foundation (Weeks 1-2)

Priority: AI Provider System Migration

Scope:

ai_provider_langchain (70% reduction)
ai_provider_apple (65% reduction)
Basic llm module provider integration (25% of total reduction)

Deliverables:

Drupal AI provider plugins for LangChain and Apple
Updated dependency management (ai:ai as required dependency)
Basic provider discovery working

Success Metrics:

Provider plugins discoverable via drush ai:providers
Existing functionality preserved
40% code reduction in provider modules

Phase 2: Core Integration (Weeks 3-5)

Priority: LLM Core Module Integration

Scope:

Complete llm module provider system migration (remaining 20% reduction)
Chat service standardization
Provider management consolidation

Deliverables:

Unified provider interface using AI module
Conversation management preserved with AI module integration
Updated service definitions leveraging contrib

Success Metrics:

All providers accessible through standard AI module interfaces
Business logic preserved (cost calculation, usage tracking)
Service count reduced by 30%

Phase 3: Agent Systems (Weeks 6-8)

Priority: AI Agent Modules

Scope:

ai_agent_crewai (45% reduction)
ai_agent_huggingface (30% reduction)
ai_agent_orchestra workflow integration

Deliverables:

Agent plugins using AI automator system
ECA workflow integration for agent orchestration
Preserved business logic with standard interfaces

Success Metrics:

Agent workflows executable via ECA
Agent discovery via AI module plugin system
35% average code reduction in agent modules

Phase 4: Platform Integration (Weeks 9-12)

Priority: Remaining Modules and Optimization

Scope:

mcp_registry (40% reduction)
gov_compliance AI logging integration (35% reduction)
Final optimization and cleanup
Documentation updates

Deliverables:

Complete ecosystem using Drupal AI module
Updated documentation and migration guides
Performance optimization and testing

Success Metrics:

All modules follow Drupal AI patterns
Overall 40% code reduction achieved
Comprehensive test coverage maintained

Technical Implementation Details

Provider Plugin Migration Pattern

Before (Custom Implementation):

// Custom provider management
class ProviderManager {
    private $providers = [];
    
    public function addProvider($name, $config) {
        $this->providers[$name] = new CustomProvider($config);
    }
    
    public function callProvider($name, $prompt) {
        return $this->providers[$name]->call($prompt);
    }
}

After (Drupal AI Plugin):

// Standard AI provider plugin
/**
 * @AiProvider(
 *   id = "langchain",
 *   label = "LangChain Provider"
 * )
 */
class LangChainProvider extends AiProviderPluginBase {
    public function call(array $messages, array $config = []): AiResponse {
        // Simplified implementation using AI module infrastructure
    }
}

Service Definition Optimization

Before (90+ custom services):

services:
  llm.provider_manager:
    class: Drupal\llm\Service\ProviderManager
    arguments: ['@config.factory', '@key.repository', '@logger.channel.llm']
  
  llm.custom_client:
    class: Drupal\llm\Service\CustomClient  
    arguments: ['@llm.provider_manager', '@logger.channel.llm']

After (Leveraging AI module):

services:
  llm.enhanced_client:
    class: Drupal\llm\Service\EnhancedClient
    arguments: ['@plugin.manager.ai_provider']  # Use AI module's provider manager
    
  # 60% of custom services eliminated by using AI module equivalents

Benefits Analysis

Code Maintainability

Reduced custom code: 35-45% less code to maintain
Community support: Leverage contrib module improvements
Standard patterns: Follow established Drupal AI conventions
Bug fixes: Benefit from community bug fixes and security updates

Performance Improvements

Optimized provider loading: AI module's lazy-loading provider system
Better caching: Built-in provider result caching
Resource management: Standardized connection pooling and cleanup

Developer Experience

Familiar APIs: Standard Drupal plugin patterns
Better documentation: AI module's comprehensive docs
IDE support: Better autocompletion and type hints
Testing: Leverage AI module's test infrastructure

Long-term Sustainability

Community alignment: Follow Drupal AI roadmap
Feature parity: Access to new AI module features automatically
Security: Centralized security updates through contrib
Interoperability: Better integration with other AI-enabled modules

Risk Assessment

Low Risk

Provider plugins: Well-established pattern, minimal breaking changes
Chat interfaces: Direct mapping to AI module equivalents
Configuration: Mostly one-to-one config mapping

Medium Risk

Agent orchestration: Complex workflows may need careful migration
Custom business logic: Ensure proper preservation during migration
Performance: Monitor response times during transition

High Risk

Government compliance: Ensure all audit trails and security features preserved
Multi-tenant isolation: Complex permission systems need careful review
Production data: Conversation history and usage data migration

Migration Validation

Testing Strategy

Unit tests: Migrate existing test coverage to new implementations
Integration tests: Verify end-to-end functionality preserved
Performance tests: Ensure no regression in response times
Security tests: Validate all compliance features maintained

Rollback Plan

Feature flags: Use configuration to switch between old/new implementations
Parallel operation: Run both systems during transition period
Data backup: Full backup of configuration and conversation data
Documentation: Detailed rollback procedures for each phase

Success Metrics

Code reduction: Achieve 35-45% reduction in custom code
Performance: Maintain or improve API response times
Functionality: 100% feature parity with existing implementations
Stability: Zero regression in existing functionality
Developer experience: Improved development workflow and debugging

Next Steps

Immediate Actions (Week 1)

Environment setup: Install Drupal AI module in development
Provider audit: Detailed analysis of ai_provider_langchain migration
Test framework: Set up parallel testing environment
Documentation: Create detailed migration checklist

Week 2-3 Deliverables

First provider plugin: Complete LangChain provider migration
Integration testing: Verify existing functionality works
Performance benchmarking: Establish baseline metrics
Team training: Developer documentation and guidelines

Ongoing Process

Weekly reviews: Progress assessment and issue resolution
Incremental deployment: Phase-by-phase production migration
Community engagement: Contribute improvements back to AI module
Knowledge sharing: Document lessons learned for future migrations

Estimated Timeline: 8-12 weeks for complete migration
Estimated Effort: 40-60 developer days across all phases
Expected ROI: 40% code reduction + improved maintainability + community support

Executive Summary​

Analysis Results​

Total Impact Assessment​

Module-by-Module Breakdown​

High Impact Migrations (55-70% reduction)​

Medium Impact Migrations (35-50% reduction)​

Lower Impact Migrations (20-35% reduction)​

Migration Strategy​

Phase 1: Foundation (Weeks 1-2)​

Phase 2: Core Integration (Weeks 3-5)​

Phase 3: Agent Systems (Weeks 6-8)​

Phase 4: Platform Integration (Weeks 9-12)​

Technical Implementation Details​

Provider Plugin Migration Pattern​

Service Definition Optimization​

Benefits Analysis​

Code Maintainability​

Performance Improvements​

Developer Experience​

Long-term Sustainability​

Risk Assessment​

Low Risk​

Medium Risk​

High Risk​

Migration Validation​

Testing Strategy​

Rollback Plan​

Success Metrics​

Next Steps​

Immediate Actions (Week 1)​

Week 2-3 Deliverables​

Ongoing Process​

Executive Summary

Analysis Results

Total Impact Assessment

Module-by-Module Breakdown

High Impact Migrations (55-70% reduction)

Medium Impact Migrations (35-50% reduction)

Lower Impact Migrations (20-35% reduction)

Migration Strategy

Phase 1: Foundation (Weeks 1-2)

Phase 2: Core Integration (Weeks 3-5)

Phase 3: Agent Systems (Weeks 6-8)

Phase 4: Platform Integration (Weeks 9-12)

Technical Implementation Details

Provider Plugin Migration Pattern

Service Definition Optimization

Benefits Analysis

Code Maintainability

Performance Improvements

Developer Experience

Long-term Sustainability

Risk Assessment

Low Risk

Medium Risk

High Risk

Migration Validation

Testing Strategy

Rollback Plan

Success Metrics

Next Steps

Immediate Actions (Week 1)

Week 2-3 Deliverables

Ongoing Process