Introducing AI-Powered Semantic Search with Meilisearch for Ghost CMS

Beyond Keywords: The Power of Understanding Meaning

Traditional search engines work on a simple principle: match the exact words a user types. This creates a fundamental disconnect:

A reader searches for "making websites fast"
Your brilliant article about "web performance optimization" doesn't appear in results
Your valuable content remains undiscovered

Semantic search changes everything. Powered by sophisticated AI models (called embedders), it understands the meaning behind both your content and search queries, not just matching keywords.

My latest update leverages Meilisearch's powerful hybrid search capabilities, combining:

Semantic Understanding: Finds posts based on conceptual similarity, even when keywords don't match
Traditional Keyword Matching: Maintains precision for exact term searches

This dual approach ensures readers find what they need, whether they know the exact terms or just the concepts they're interested in.

What's New in the Search Experience?

When you enable AI Search in the Ghost-Meilisearch integration, your readers will see a refined search interface:

AI Suggestions: A new section at the top displays the most semantically relevant results based on the meaning of the search query. This is perfect for discovering related concepts and exploring topics more broadly.
Keyword Matches: Below the AI suggestions, the familiar list of results based on direct keyword matches remains, ensuring users can still find specific terms if needed.

This two-pronged approach provides a richer, more intuitive discovery experience, helping readers uncover more of your great content.

Getting Started with AI Search

Ready to bring semantic search to your Ghost blog? Here's what you need:

Prerequisites:

An Updated Meilisearch Instance: The AI features require a recent version of Meilisearch that supports embeddings.
A Configured Embedder: You need to have an embedder configured within your Meilisearch instance settings. Note the name you give your embedder.

For example, if you're using OpenAI as your embedder, you would configure it with a command like this:

curl \
  -X PATCH 'MEILISEARCH_URL/indexes/indexes/ghost_posts/settings/embedders' \
  -H 'Content-Type: application/json' \
  --data-binary '{
    "ghost-posts-openai": {
      "source": "openAi",
      "apiKey": "OPEN_AI_API_KEY",
      "model": "text-embedding-3-small",
      "documentTemplate": "Blog post titled '{{ doc.title }}' with excerpt as '{{ doc.excerpt }}'"
    }
  }'

Setup Steps:

Update Search UI Assets in Ghost: Go to Settings → Code injection in your Ghost Admin:
- Replace "your-embedder-name" with the actual name you configured in Meilisearch.
- Set enableAiSearch to true.
Configure AI Search: Still in the Site Header code injection, find or add your window.__MS_SEARCH_CONFIG__ block and add the new AI options:

<script>
  window.__MS_SEARCH_CONFIG__ = {
    // --- Your existing config ---
    meilisearchHost: "https://your-meilisearch-host.com",
    meilisearchApiKey: "your-search-only-api-key",
    indexName: "ghost_posts", // Or your custom index name
    theme: "system", // Optional: 'light', 'dark', or 'system'

    // --- New AI Search Configuration ---
    enableAiSearch: true, // Set to true to turn it on!
    aiSearchEmbedder: "your-embedder-name", // Replace with the name of your configured Meilisearch embedder
    aiSearchLimit: 3 // Optional: Max results in "AI Suggestions" (default: 3)
  };
</script>

Update the Site Header script URL to:

<script src="https://cdn.jsdelivr.net/npm/@fanyangmeng/ghost-meilisearch-search-ui@1.0.1/dist/search.min.js"></script>

Self-hosters: If you're hosting the search UI script yourself or using the config.[environment].json method, update the script URL and add the AI configuration options accordingly.

"sodoSearch": {
    "url": "https://cdn.jsdelivr.net/npm/@fanyangmeng/ghost-meilisearch-search-ui@1.0.1/dist/search.min.js"
  }

Update the CSS: Still in the Site Header code injection, simply add this line to it:

<link rel="stylesheet" type="text/css" href="https://cdn.jsdelivr.net/npm/@fanyangmeng/ghost-meilisearch-search-ui@1.0.1/dist/styles.css">

Why AI-Powered Search Matters for Your Blog

Understanding the value of this update goes beyond the technical implementation. AI-powered search represents a fundamental improvement in how readers interact with your content:

Greater Content Discovery With semantic search, readers are more likely to find your valuable posts, even if they search using different terminology than you used in your writing. This means more eyes on your content and less valuable information going undiscovered.

Reduced Search Frustration We've all experienced the frustration of knowing information exists somewhere but not being able to find it because we don't know the right search terms. Semantic search eliminates this problem by focusing on concepts rather than specific keywords.

Deeper Engagement with Your Content When readers can easily find related content based on concepts rather than just keywords, they're likely to explore more of your blog and engage more deeply with your ideas. This can lead to longer session times, more page views, and a more satisfied audience.

A Premium Search Experience By implementing AI-powered search, you're offering your readers the same sophisticated search capabilities found on major publications and platforms. This elevates the perceived quality of your blog and improves the overall user experience.

Looking Forward: The Evolution of Blog Search

The introduction of AI-powered semantic search represents a significant step forward in how readers discover and interact with blog content. By bridging the gap between what readers are thinking and the specific words used in your posts, you're creating a more intuitive, satisfying experience for your audience.

As this technology continues to evolve, we can expect even more sophisticated understanding of content and search intent, further narrowing the gap between what readers want to find and what they can easily discover.

Getting Started Today

I encourage you to give Ghost-Meilisearch v1.0.1 with AI-powered search a try on your blog. The setup process may take a little technical effort, but the improved reader experience is well worth it.

After implementing the feature:

Test it thoroughly with different types of searches to see how it performs
Monitor how readers interact with the new search experience
Consider highlighting the improved search capability to your regular readers

I am excited to hear about your experience with this new feature! The complete project documentation and support resources are available on GitHub. If you have questions, suggestions, or feedback, please open an issue in the repository—your input will help shape the future development of this integration.

By embracing AI-powered semantic search, you're not just upgrading a technical component of your blog; you're fundamentally improving how readers discover and engage with your valuable content. That's something worth celebrating!