Unstructured API key

Time to complete: ~2 minutes

The unstructured API key gives you access to Read (text analysis) and Vision (image analysis). It is separate from per-model keys used by ML, Forecast, and Loop.

Before you start

• Heimdall account
• You need Read/Vision only — skip this page if you are doing tabular ML via Lake

How to get your key

1. Sign in at heimdallapp.org
2. Either:
   • Sidebar footer → Looking for unstructured? (opens Settings), or
   • Sidebar → Account → API keys (/settings?section=api-keys)
3. Click Generate under the unstructured key section

Important: Store Your API Key Securely

Unstructured API key is issued one time and cannot be retrieved. If you lose your API key, you'll need to generate a new one and update your applications.

What You Can Do with Your Unstructured API Key

Your single unstructured API key gives you access to:

• Text Analysis - Extract sentiment, topics, and insights from text data
• Image Analysis - Analyze images and extract features

Using Your Unstructured API Key

Include your unstructured API key in the headers of all API requests for both text and image analysis:

headers = {
    'X-api-key': 'YOUR-UNSTRUCTURED-API-KEY',
    'X-username': 'YOUR-USERNAME'
}

Expected Response

When you analyze text data, you'll receive a comprehensive analysis like this:

{
  "length": 108,
  "word_count": 17,
  "oov_ratio": 0.17647058823529413,
  "oov_ratio_2": 0.29411764705882354,
  "sentence_count": 1,
  "avg_word_length": 5.470588235294118,
  "avg_sentence_length": 17,
  "noun_count": 7,
  "verb_count": 3,
  "adjective_count": 2,
  "adverb_count": 0,
  "pronoun_count": 0,
  "stopword_count": 5,
  "tfidf_top1": "accessible",
  "tfidf_top2": "amazing",
  "tfidf_top3": "data",
  "sentiment": "positive",
  "compound_sentiment_score": 0.5859
}

Security Best Practices

• Never commit API keys to version control
• Use environment variables to store API keys
• Rotate keys regularly for security

Lake vs unstructured key

Need	Use
One text/image per API call	This key → Read/Vision
Thousands of files for ML training	Lake → Add data → Unstructured zip — no unstructured key required for ingest

Next steps

1. First API call
2. Unstructured suite overview
3. Tabular ML? → User journeys (Lake path, not this key)