Process and analyze data with AI.
Text Summarization
Condense long documents:
python
1from inferencesh import inference2 3client = inference(api_key="inf_your_key")4 5result = client.run({6 "app": "infsh/summarize",7 "input": {8 "text": long_document,9 "max_length": 20010 }11})12 13print(result["output"]["summary"])Document Analysis
Extract information from documents:
python
1result = client.run({2 "app": "infsh/document-qa",3 "input": {4 "document": "/path/to/contract.pdf",5 "questions": [6 "What is the contract duration?",7 "What are the payment terms?",8 "Who are the parties involved?"9 ]10 }11})12 13for answer in result["output"]["answers"]:14 print(f"Q: {answer['question']}")15 print(f"A: {answer['answer']}\n")Sentiment Analysis
Analyze text sentiment:
python
1result = client.run({2 "app": "infsh/sentiment",3 "input": {4 "texts": [5 "This product is amazing!",6 "Terrible customer service.",7 "It's okay, nothing special."8 ]9 }10})11 12for item in result["output"]["results"]:13 print(f"{item['text'][:30]}... → {item['sentiment']} ({item['score']:.2f})")Output:
code
1This product is amazing!... → positive (0.95)2Terrible customer service.... → negative (0.89)3It's okay, nothing special.... → neutral (0.67)Batch Processing
Process many items efficiently:
python
1texts = [...] # Your data2 3# Run in batches4batch_size = 105results = []6 7for i in range(0, len(texts), batch_size):8 batch = texts[i:i + batch_size]9 10 result = client.run({11 "app": "infsh/summarize",12 "input": {"texts": batch}13 })14 15 results.extend(result["output"]["summaries"])With Agent
Create a data analysis agent:
code
1You: Analyze this CSV file and tell me the key insights2 3Agent: I'll analyze your data.4 [Loading CSV...]5 [Running analysis...]6 7 Key insights from your sales data:8 9 �� Overview:10 - Total records: 15,43211 - Date range: Jan 2023 - Dec 202412 - Total revenue: $2.4M13 14 �� Trends:15 - Q4 shows 23% higher sales than Q116 - Product A is top performer (34% of revenue)17 - Customer retention rate: 67%18 19 ⚠️ Anomalies:20 - March 2024 shows unusual spike (investigate)21 - Category C declining (-15% YoY)22 23 Would you like me to create visualizations?Common Data Apps
| App | Use Case |
|---|---|
summarize | Condense long text |
sentiment | Analyze tone/sentiment |
document-qa | Q&A over documents |
extract | Extract structured data |
translate | Language translation |
classify | Categorize text |
Tips
For large datasets:
- Use batch processing
- Consider streaming for real-time results
- Run on private workers for data privacy
For accuracy:
- Clean your data first
- Be specific in prompts
- Validate results on a sample