Splabs

This product analyzes AI behavior from text output alone, detecting issues like drift, sycophancy, and hallucination risk without needing model access. It serves developers, trust and safety teams, and enterprise compliance, targeting organizations deploying large language models. It is delivered as a SaaS platform with an API and web app, offering deterministic, auditable scoring and privacy-safe forensic records.

Key features

• Behavioral classification with 13 classifiers
• 116 behavioral classes and 38 deterministic metrics
• Dyadic Sequence Analysis for one-to-one conversations
• Agent Swarm Analysis for group behavior contagion
• Summary Distortion Analysis for meaning changes
• Retrieval Drift Monitor for RAG bias detection
• Distributed Exfiltration Detection for secret leaks
• Psychological risk profile with 100 indicators
• Privacy-safe incident archive with GDPR erasure
• Real-time alerts for risky conversations
• Auditable named-rule scoring
• Supports 5 languages