• Classification
  • Generation
  • Web Enrichment
  • Geolocation Enrichment
  • File Extraction
  • File Extraction (OCRs)

​
Model Selection & Credits

We offer the best AI models on the market, with the additional possibility to leverage Claro’s proprietary model that also generates confidence scores and citations.

​
Model Pricing

Task pricing varies based on the AI model used.

​
Classification

Automatically categorize and tag your data with consistent labels.

How it works: Compares each row against your predefined categories or asks AI to suggest classifications based on patterns in your data.

✅ Perfect for:

  • Product taxonomy: “Electronics → Laptops → Gaming Laptops”
  • Customer support: Route tickets to Bug/Feature/Billing teams
  • Content moderation: Flag positive/neutral/negative sentiment
  • Lead scoring: Qualify prospects as hot/warm/cold

⚠️ Limitations:

  • Works best with ≤50 categories for accuracy
  • Very nuanced distinctions may need human review

💡 Optimization tips:

  • Start with 10-20 categories, expand gradually
  • Provide 2-3 examples per category
  • Adjust confidence threshold in settings (default: 60)

​
Generation

Create or rewrite text content tailored to your specifications.

How it works: Uses advanced language models to generate new content based on your prompts and existing data.

✅ Perfect for:

  • SEO content: Product descriptions from specifications
  • Sales outreach: Personalized emails from prospect data
  • Marketing copy: Social media posts from product features
  • Documentation: Summaries from technical specifications

⚠️ Limitations:

  • Avoid for legal documents or contracts
  • May require human review for brand voice consistency

💡 Optimization tips:

  • Specify exact length: “Write exactly 150 characters.”
  • Include brand voice guidelines in the Knowledge Base
  • Use temperature settings for creativity vs. consistency

​
Web Enrichment

Research live information from across the web to enhance your data.

How it works: Searches the web for relevant information, extracts key details, and provides citations for verification.

✅ Perfect for:

  • Competitive intelligence: Pricing, features, positioning
  • Company research: Funding, employees, recent news
  • Product validation: Reviews, specifications, availability
  • Contact enrichment: Social profiles, company details

⚠️ Limitations:

  • Cannot access paywalled or private content
  • Results depend on publicly available information
  • It may be slower due to web scraping

💡 Optimization tips:

  • Keep prompts specific and focused
  • Always request citations for traceability
  • Best results with well-known entities/brands

​
Geolocation Enrichment

Transform addresses into rich geographic data and nearby points of interest.

How it works: Converts addresses to coordinates and optionally finds surrounding businesses, amenities, or services.

✅ Perfect for:

  • Store locators: Add coordinates and nearby competitors
  • Customer segmentation: Urban vs. rural classification
  • Logistics optimization: Route planning and territory mapping
  • Market analysis: Demographics and local business density

⚠️ Limitations:

  • Less accurate for rural or remote locations
  • P.O. boxes cannot be geocoded

💡 Optimization tips:

  • Use tight radius (1-2km) in urban areas
  • Chain with classification for location types
  • Validate important addresses manually

​
File Extraction

Extract structured data from PDFs, invoices, forms, and documents.

How it works: AI-powered document parser that identifies and extracts specific fields without templates.

✅ Perfect for:

  • Invoice processing: Vendor, amount, due date, line items
  • Form digitization: Applications, surveys, registrations
  • Compliance documents: Safety data sheets, certifications
  • Contract analysis: Key terms, dates, parties

⚠️ Limitations:

  • Works best with standard document layouts
  • Handwritten content needs Advanced OCR instead

💡 Optimization tips:

  • Specify the exact fields you need in the prompt
  • Use consistent file naming for batch processing
  • Store document templates in the Knowledge Base

​
Advanced OCR - on request (4 credits)

Extract text from low-quality, handwritten, or complex visual documents.

How it works: Vision-AI pipeline optimized for challenging documents that standard OCR cannot handle.

✅ Perfect for:

  • Handwritten forms: Applications, surveys, notes
  • Damaged documents: Faded receipts, torn papers
  • Skewed scans: Photos of documents, angled captures
  • Complex layouts: Multi-column, mixed media

⚠️ Limitations:

  • Most expensive option (8 credits)
  • Processing time is longer than other tasks
  • Very complex layouts may need preprocessing

💡 Optimization tips:

  • Try File Extraction first for standard documents
  • Improve image quality when possible
  • Break complex documents into sections

​
Setting Up AI Tasks

Writing Effective Prompts:

  • Reference other columns with @column_name syntax: “Based on @company_name and @industry, write a personalized cold email introduction”
  • Be specific about format and length: “Classify @product_title into exactly one category: Electronics, Clothing, Home, Books, or Other”
  • Include examples when helpful: “Extract the total amount from @invoice_file. Return only the number like: 1,234.56”

Using Knowledge Bases: Attach relevant documents to improve accuracy:

  • Product manuals for technical specifications
  • Brand guidelines for a consistent voice
  • Policy documents for compliance checks
  • Industry standards for classification

Best practices:

  • Keep Knowledge Bases focused and relevant
  • Update documents regularly
  • Use descriptive names for easy selection

Model Selection Tips:

  • Claro Default: Best value with confidence scoring and citations included
  • Other models: Excellent for specialized use cases with variable pricing