Open Intranet + Milvus RAG Demo

Integration demo of Open Intranet (Drupal 11) with Milvus Vector Database for RAG (Retrieval-Augmented Generation).

The main Open Intranet repository and project is available at https://www.drupal.org/project/openintranet.

Prerequisites

Before running the installation script, ensure you have:

1. Docker Desktop - Running and active
2. DDEV - Installed (brew install ddev/ddev/ddev on macOS)
3. OpenAI API Key - Required for generating embeddings
   • Get it from: https://platform.openai.com/api-keys
   • Key must start with sk-proj- or sk-
   • Billing must be enabled (~$0.01-0.10 for the entire demo)

Installation

git clone https://github.com/droptica/openintranet_rag_demo.git
cd openintranet_rag_demo
./launch_openintranet_with_rag_demo.sh

The script will:

1. Clone Open Intranet from Drupal.org
2. Download Milvus docker-compose configuration
3. Configure DDEV (Drupal 11, PHP 8.3)
4. Start containers (web, db, Milvus)
5. Install Composer dependencies
6. Copy DDEV commands and starter theme
7. Add drupal/ai_vdb_provider_milvus:^1.1@beta module
8. Copy openintranet_milvus_rag recipe
9. Install Drupal with demo content
10. Apply Milvus RAG recipe configuration
11. Prompt interactively for OpenAI API key (validates format)
12. Save API key to Drupal's Key module
13. Index Knowledge Base content to Milvus
14. Generate one-time login URL

During installation, you'll be prompted to paste your OpenAI API key. The script validates the format and stores it securely.

Structure

openintranet_rag_demo/
├── launch_openintranet_with_rag_demo.sh  # Main setup script
├── recipes/
│   └── openintranet_milvus_rag/          # Drupal recipe for Milvus integration
│       ├── recipe.yml
│       ├── composer.json
│       └── config/
│           ├── ai_provider_openai.settings.yml
│           ├── ai_vdb_provider_milvus.settings.yml
│           ├── search_api.server.milvus_embeddings.yml
│           ├── search_api.index.knowledge_base_content.yml
│           ├── ai_search.index.knowledge_base_content.yml
│           ├── key.key.openai_api_key.yml
│           └── views.view.rag_search_example.yml
└── openintranet_source_code/
    └── openintranet/                      # Cloned Open Intranet repository
        ├── composer.json
        ├── ddev_commands/                 # Custom DDEV commands
        ├── recipes/                       # Additional Drupal recipes
        ├── starter-theme/                 # OpenIntranet theme
        └── web/                           # Drupal root

What it installs

• Open Intranet: Drupal starter kit for intranet sites
• Milvus VDB Provider: drupal/ai_vdb_provider_milvus:^1.1@beta
• Milvus: Vector database (via docker-compose)
• RAG Search Example: Search page at /search-rag-example using Milvus embeddings

Verification

After installation, verify everything is working:

Check Search Index Status

cd openintranet_source_code/openintranet
ddev drush search-api:status

Expected output:

knowledge_base_content   Knowledge Base Content   100%         24        24

If you see 100% - everything is working! 🎉

Verify Milvus Connection

1. Open Milvus Attu UI: Check the port with ddev describe (look for Attu service port)
2. Connect to: http://milvus:19530
3. Find collection: openintranet_knowledge_base
4. Check: Entity Count > 0 ✅

Test OpenAI API

cd openintranet_source_code/openintranet
ddev drush php:eval "
\$provider = \Drupal::service('ai.provider')->createInstance('openai');
\$result = \$provider->embeddings('test', 'text-embedding-3-small', []);
echo count(\$result->getNormalized()) . ' dimensions';
"

Expected: 1536 dimensions

Testing RAG Search

The recipe includes a RAG Search Example page at /search-rag-example. To test:

1. Access the page: https://your-site.ddev.site/search-rag-example
2. Enter a search query (e.g., "milvus configuration")
3. Verify results display with:
   • Title (linked to source page)
   • Content excerpt
   • Relevance score

Verify View Installation

cd openintranet_source_code/openintranet
ddev drush views:list | grep rag_search

Expected output:

rag_search_example   RAG Search Example   Page    Enabled

Troubleshooting

Invalid API key format

Issue: Script says "Invalid API key format"

Solution:

• Ensure the key starts with sk- or sk-proj-
• Copy the entire key including the prefix
• Get the key from https://platform.openai.com/api-keys

Failed to save API key

Solution:

cd openintranet_source_code/openintranet
ddev drush config:set key.key.openai_api_key key_provider_settings.key_value "sk-proj-YOUR_KEY" -y
ddev drush config:set ai_provider_openai.settings api_key openai_api_key -y
ddev drush cache:rebuild
ddev drush search-api:index knowledge_base_content

Entity Count = 0 in Milvus

Solution:

cd openintranet_source_code/openintranet
# Test API key first (should return 1536 dimensions)
ddev drush php:eval "
\$provider = \Drupal::service('ai.provider')->createInstance('openai');
\$result = \$provider->embeddings('test', 'text-embedding-3-small', []);
echo count(\$result->getNormalized()) . ' dimensions';
"

# If API key works, reindex
ddev drush search-api:clear knowledge_base_content
ddev drush search-api:index knowledge_base_content

No search results

Solution:

cd openintranet_source_code/openintranet
# Reindex everything
ddev drush search-api:clear knowledge_base_content
ddev drush search-api:reindex knowledge_base_content
ddev drush search-api:index knowledge_base_content
ddev drush cache:rebuild

Page not found (404) for search page

Solution:

cd openintranet_source_code/openintranet
ddev drush cache:rebuild
ddev drush router:rebuild

Debugging Commands

View errors

ddev drush watchdog:show --severity=Error --count=20

Check Milvus connection

ddev describe
ddev exec curl -I http://milvus:19530

Reindex content

ddev drush search-api:clear
ddev drush search-api:reindex
ddev drush search-api:index

---

About Droptica

Built with ❤️ by Droptica 🇵🇱

Solid Open Source solutions for ambitious companies.

What we do:

• Create: Open Intranet, Droopler CMS, Druscan
• AI Development: AI chatbots (RAG), autonomous agents, OpenAI/Claude integrations, custom AI models, CMS content automation & translation, workflow automation
• Customize: Drupal, Mautic, Sylius, Symfony
• Support & maintain: Security, updates, training, monitoring 24/7

Trusted by: Corporations • SMEs • Startups • Universities • Government