Research
Research applied to real problems. For people who need to understand, invest and act, beyond the research community.
Retrieval Is Not Enough: Why Organizational AI Needs Epistemic Infrastructure
The OIDA paper: a framework that assigns computable epistemic state to organisational knowledge, with public evaluation corpora.
Read the sourceSummary Is Not Enough: Source-Blind LLM Judges Mistake Faithful Citation for Hallucination
LLM judges without access to sources mistake faithful citations for model errors.
Read the sourceAssessing AI Act Compliance: An LLM Tool for Enterprises
A tool that maps compliance requirements across the AI Act, GDPR, DSA and DMA.
Read the sourceA Multi-Layer AI Framework for Information Landscape Analysis
An architecture integrating strategic reasoning, automation and collaboration between people and AI.
Read the sourceBeyond Sentiment: Why Traditional NLP Fails Political News, and How LLMs Can Bridge the Gap
A study of “neutral collapse” in sentiment analysis of political discourse.
Read the sourceBloc-Conditional Event States: Measuring Cross-Coverage Divergence for Threat-Intelligence Analysis
An open-source threat intelligence method based on divergence in media coverage.
Read the sourceMAGELLAN: Autonomous Cross-Disciplinary Scientific Discovery via Multi-Agent Adversarial Scrutiny
A fifteen-agent system connecting knowledge across scientific domains through adversarial validation.
Read the source
Research conducted at Kakashi Venture Accelerator, the venture studio that accelerated OIDA.