QuerySphere

Break Down Information Silos Instantly

Transform how your organization accesses and utilizes knowledge across documents, folders, and archives with AI-powered search.

Upload Documents

Upload your documents to build an intelligent knowledge base

Drag & drop files here

or click to browse your computer

Supports PDF, DOCX, TXT, ZIP • Max 2GB per file

No files uploaded yet

Chat with Documents

Ask questions about your knowledge base

Press Enter to send • Shift+Enter for new line

Analytics & Quality Metrics

Session Statistics

Total Evaluations

Average Answer Relevancy

Average Faithfulness

Average Context Utilization

RAGAS Evaluation History (0 entries)

#	Timestamp	Query	Answer	Query Type	Answer Relevancy	Faithfulness	Context Precision	Context Relevancy	Retrieval Time (ms)	Generation Time (ms)	Total Time (ms)	Chunks Retrieved
No RAGAS evaluations yet. Start chatting to see quality metrics.

Metric Descriptions:

Query Type: RAG (uses documents) vs GENERAL (no documents)

Answer Relevancy: How well the answer addresses the question (0-1)

Faithfulness: Is the answer grounded in the retrieved context (0-1)

Context Utilization: How well the retrieved context is utilized (0-1)

Color coding: ≥0.7 (Good), 0.4-0.7 (Fair), <0.4 (Poor), N/A (Not Applicable)

Table automatically refreshes when you switch to this tab or send a new message

System Configuration

Inference Model

Select the LLM provider and model for text generation

Chunking Strategy

Choose how documents are split into chunks for processing

Chunking Parameters

Chunk Size (tokens)

Larger chunks capture more context but may reduce precision

Chunk Overlap (tokens)

Overlap between chunks to maintain context continuity

Retrieval Parameters

Retrieval Top K

Number of chunks to retrieve for each query

Vector Weight

Weight for vector similarity search (0.0-1.0)

BM25 Weight

Weight for keyword search (0.0-1.0)

Enable Cross-Encoder Reranking

Use transformer model to rerank results for better precision (slower but more accurate)

Generation Parameters

Temperature

Lower for factual answers, higher for creative responses

Max Tokens

Maximum length of generated responses

Current System Status

LLM Health: -

Vector Store: -

Embeddings: -

Retrieval: -

Generation: -

Overall Status: -

Break Down Information Silos Instantly

Multi-Format Support

Intelligent Processing

Hybrid Retrieval

Smart Q&A

Quality Analytics

Flexible Configuration

Get Started in 4 Simple Steps

Upload Your Content

Build Knowledge Base

Chat with AI Assistant

Optimize & Export

Document Upload