PDF, Word, PowerPoint, Excel, TXT and Markdown files are supported, with OCR for scanned pages, bulk folder import and automatic duplicate detection.
Automated knowledge construction
AiKMSpark reads each document and extracts summaries, attributes, entities and relationships, requiring no predefined taxonomy.
Reviewable and editable knowledge base
Entries can be reviewed, corrected, and merged where a single subject spans multiple records.
Traceable answers and reports
Multi-turn conversational Q&A returns cited sources, showing which knowledge entry and field matched each question, and findings can be compiled into reports.
On-premises deployment
The system runs on one machine and is accessed through a browser; models operate offline, with cloud API connectivity optional.