distillcore
Skip to Content
Presets

Presets

Presets are pre-configured DomainConfig objects that provide domain-specific LLM prompts for classification, structuring, and enrichment.

Available Presets

from distillcore import list_presets
 
print(list_presets())  # ['generic', 'legal']

Generic Preset

The default preset for general-purpose documents.

from distillcore import load_preset
 
config = load_preset("generic")

Classification extracts: document_type, document_title, author, date, summary

Structuring sections: title, header, body, conclusion, appendix, and hierarchical sub-sections

Enrichment per chunk: topic, key_concepts, relevance score

Optimized for legal documents including motions, orders, transcripts, and depositions.

config = load_preset("legal")

Classification extracts: document_type, filing_party, case_number, court, judge, filing_date, is_transcript, attorneys

Structuring sections: argument, holding, opinion, order, and legal document sections

Transcript support: Automatically detects depositions and court transcripts, parsing speaker turns with Q/A formatting.

Using a Preset

from distillcore import process_document, DistillConfig, load_preset
 
config = DistillConfig(
    domain=load_preset("legal"),
    openai_api_key="sk-...",
)
 
result = process_document("motion.pdf", config=config)
print(result.document.metadata)  # includes case_number, court, etc.

Creating a Custom Preset

Implement a DomainConfig with your own prompts:

from distillcore import DomainConfig
 
medical_preset = DomainConfig(
    name="medical",
    classification_prompt="Analyze this medical document...",
    structuring_prompt="Extract sections from this clinical note...",
    transcript_prompt="Parse this patient consultation...",
    enrichment_prompt="For this medical text chunk...",
    parse_classification=my_custom_parser,  # optional
)

Then use it:

config = DistillConfig(domain=medical_preset)
result = process_document("clinical_note.pdf", config=config)
ConfigurationExtractors