Semantic Search

Search your brain by meaning, not just keywords. NeuralSnap converts your query into a vector embedding and finds the most semantically similar snapshots using cosine similarity — so searching for "dealing with uncertainty" will surface snapshots about ambiguity tolerance, probabilistic thinking, and comfort with the unknown, even if they never use the word "uncertainty".

How It Works

Query text→Embedding→Cosine similarity→Ranked results

Your query is converted into a high-dimensional vector, compared against all snapshot embeddings in your brain, and results are returned ranked by semantic similarity.

Search Snapshots

GET/api/v1/snapshots/search

Search your snapshots by semantic similarity.

Query Parameters

qstringrequired

The search query. Can be a word, phrase, question, or full sentence.

limitnumberdefault: 10

Maximum number of results to return (1–100).

brain_iduuid

Optional brain ID to restrict search scope.

bash

curl "https://neuralsnap.ai/api/v1/snapshots/search?q=dealing%20with%20uncertainty&limit=5" \
  -H "Authorization: Bearer ns_test_abc123"

json

{
  "data": [
    {
      "id": "snap_x9y8z7w6v5u4",
      "brain_id": "brain_abc123",
      "name": "Embrace Ambiguity",
      "core": "The best decisions are often made with incomplete information",
      "type": "Conviction",
      "one_liner": "Comfort with not-knowing is a competitive advantage",
      "confidence": 0.85,
      "significance_score": 0.8,
      "tags": ["decision-making", "mindset"],
      "similarity": "0.9230",
      "created_at": "2025-01-10T12:00:00.000Z"
    },
    {
      "id": "snap_m3n4o5p6q7r8",
      "brain_id": "brain_abc123",
      "name": "Probabilistic Thinking",
      "core": "Think in probability distributions, not binary outcomes",
      "type": "Model",
      "one_liner": "Nothing is 100% — assign likelihoods and update them",
      "confidence": 0.9,
      "significance_score": 0.75,
      "tags": ["thinking", "probability", "decision-making"],
      "similarity": "0.8710",
      "created_at": "2025-01-08T09:30:00.000Z"
    }
  ],
  "count": 2,
  "query": "dealing with uncertainty",
  "mode": "semantic",
  "durationMs": 312
}

Embedding fallback

If vector embeddings are temporarily unavailable, the API automatically falls back to text-based matching (keyword search) and returns results without the similarity field. Results will still be relevant but may be less precise.

Try it — Search your brain

q

limit

Snapshots

Crystallize