foundation models

QORPUS, our flagship product, uses large language models from multiple providers including Anthropic (Claude), OpenAI (GPT), and Google (Gemini). These models generate text-based outputs in response to user prompts, structured through our divergence-as-signal methodology which compares outputs across models to surface disagreement and improve reliability.

human oversight

All AI-generated outputs in QORPUS are subject to human review. Our BLOOMING CONSCIENCE agent system provides automated verification, claim checking, and bias detection before outputs reach the user. No autonomous decisions are made without human confirmation. The system is designed to augment human judgement, not replace it.

known limitations

Large language models can produce incorrect, biased, or misleading outputs. They may hallucinate facts, reflect training data biases, or fail on tasks requiring real-time information. QORPUS mitigates these risks through multi-model comparison and agent-based verification, but no AI system is infallible. Users should apply their own judgement to all outputs.