Features
Sentiment Analysis (positive/negative/neutral)
Twitter Sentiment Analysis
Subjectivity Analysis (subjective/objective)
Topic Classification (12 categories)
Readability Assessment (basic/intermediate/advanced)
Language Detection (96 languages)
Text Extraction from webpages
Document Similarity estimation
REST API with JSON responses
Open-source Java ML framework
Evaluate AI agents with objective metrics (groundedness, context relevance, coherence, etc.)
Trace agent execution flow via OpenTelemetry
Measure groundedness, context relevance, answer relevance, coherence, comprehensiveness
Detect harmful or toxic language, user sentiment, language mismatch, fairness, bias
Create custom evaluation metrics
Compare LLM app versions on a metrics leaderboard
Identify trace-level regressions across versions
Ingest OpenTelemetry traces from any agent
Python SDK for instrumenting agents
Blocking guardrails for real-time content filtering
Supports agents, RAG, summarization, copilots
