Quick Start · MongoDB Search And AI · mongodb/agent-skills · Skills Docs

Quick Start · MongoDB Search And AI · mongodb/agent-skills · Skills Docs

Here’s how Automated Embedding works:

MongoDB handles everything — you don’t write any embedding code. When you create this index, MongoDB reads every movie plot and converts it into a vector (a list of numbers that captures its meaning) using a Voyage AI model called voyage-4. It does this automatically, in the background, every time you add or update a document.

Token allocation:

You get 200 million free tokens per model, per organization — one-time, shared across all your Atlas projects

After that, voyage-4 costs $0.06 per million tokens

For reference: ~21,000 movie plots is roughly 5–10M tokens total for the initial sync

If you’re on M0: Once your 200M free tokens are exhausted, MongoDB automatically invoices you for additional usage — index builds and queries don’t stop. You can add a payment method to your Atlas account without upgrading your cluster (Atlas → Billing → Payment Method). Charges are only for embedding model usage.

Two more things worth knowing before you commit:

Where the embedding happens. The embedding model runs on inference infrastructure that MongoDB operates — a multi-tenant service on Google Cloud in a US region — regardless of which cloud provider or region your cluster is in. Your text is sent there to be embedded. Data transfer costs apply on top of the token costs above. If you have data-residency requirements, this is the detail to check first.

Where the embeddings are stored. They live on your cluster, in a dedicated internal database (one generated-embeddings collection per autoEmbed index). So they consume your disk, which is why dedicated clusters need storage auto-scaling.

The index we’d create looks like this:
{
  "fields": [
    { "type": "autoEmbed", "modality": "text", "path": "plot", "model": "voyage-4" },
    { "type": "filter", "path": "genres" }
  ]
}

db.movies.aggregate([
  {
    $vectorSearch: {
      index: "quickstart_semantic",
      path: "plot",
      query: "a story about growing up",
      numCandidates: 100,
      limit: 3
    }
  },
  {
    $project: {
      _id: 0,
      title: 1,
      plot: 1,
      score: { $meta: "vectorSearchScore" }
    }
  }
])

db.movies.aggregate([
  {
    $vectorSearch: {
      index: "quickstart_semantic",
      path: "plot",
      query: "a story about growing up",
      filter: { genres: { $eq: "Drama" } },
      numCandidates: 150,
      limit: 3
    }
  },
  {
    $project: {
      _id: 0,
      title: 1,
      genres: 1,
      score: { $meta: "vectorSearchScore" }
    }
  }
])

db.embedded_movies.aggregate([
  {
    $vectorSearch: {
      index: "quickstart_manual",
      path: "plot_embedding",
      queryVector: <the plot_embedding array fetched for "Toy Story" above>,
      numCandidates: 100,
      limit: 5
    }
  },
  {
    $project: {
      _id: 0,
      title: 1,
      plot: 1,
      score: { $meta: "vectorSearchScore" }
    }
  }
])

db.movies.aggregate([
  {
    $search: {
      index: "quickstart_text",
      text: {
        query: "space adventure",
        path: ["title", "plot"]
      }
    }
  },
  {
    $project: {
      _id: 0,
      title: 1,
      plot: 1,
      score: { $meta: "searchScore" }
    }
  },
  { $limit: 3 }
])

db.movies.aggregate([
  {
    $search: {
      index: "quickstart_text",
      text: {
        query: "sapce adventre",
        path: ["title", "plot"],
        fuzzy: { maxEdits: 1 }
      }
    }
  },
  {
    $project: { _id: 0, title: 1, score: { $meta: "searchScore" } }
  },
  { $limit: 3 }
])

db.movies.aggregate([
  {
    $search: {
      index: "quickstart_autocomplete",
      autocomplete: { query: "inc", path: "title" }
    }
  },
  {
    $project: { _id: 0, title: 1 }
  },
  { $limit: 5 }
])

db.movies.aggregate([
  {
    $rankFusion: {
      input: {
        pipelines: {
          semanticPipeline: [
            {
              $vectorSearch: {
                index: "quickstart_semantic",
                path: "plot",
                query: "a story about growing up",
                numCandidates: 100,
                limit: 20
              }
            }
          ],
          keywordPipeline: [
            {
              $search: {
                index: "quickstart_text",
                text: { query: "coming of age", path: "plot" }
              }
            },
            { $limit: 20 }
          ]
        }
      },
      combination: {
        weights: {
          semanticPipeline: 0.7,
          keywordPipeline: 0.3
        }
      }
    }
  },
  {
    $project: { _id: 0, title: 1, plot: 1 }
  },
  { $limit: 5 }
])