SIGI exists to build the scientific foundation for understanding how generative AI systems select, cite, and surface information. We believe that as AI-generated responses replace traditional search results for millions of queries, the mechanisms behind source selection deserve rigorous, independent study.
Our research spans LLM citation behaviour, generative search patterns, content architecture for AI readability, and the emerging discipline of Generative Engine Optimization. We publish our findings openly and develop reproducible methodologies that meet rigorous research standards of evidence.
As AI increasingly mediates access to information, we work to ensure the processes behind that mediation are understood, documented, and transparent.
How do large language models decide which sources to cite? We study the patterns, biases, and determinants of citation in AI-generated responses across all major platforms.
The emerging science of structuring content for AI readability. We research what makes content more likely to be surfaced, cited, and accurately represented by LLMs.
Developing standardised, reproducible research methodologies for studying generative AI behaviour that meet rigorous research standards and enable cross-study comparison.
SIGI operates as the dedicated research arm of TDS Australia. We believe transparency about this relationship is essential to our credibility. TDS provides operational support, infrastructure, and funding. In return, SIGI's research informs the broader TDS ecosystem.
However, SIGI's research agenda is set independently. Our methodology is published openly, our data collection processes are documented, and our findings are reported regardless of whether they support commercial applications.
The TDS ecosystem includes commercial services such as TDS DaaS (Design as a Service) and TDS GEO Agency (applied Generative Engine Optimization). SIGI's research provides the scientific foundation for these services, but does not direct their commercial operations.
We publish all conflicts of interest alongside our research and welcome scrutiny of our methodology and findings.
SIGI papers are self-published original research. They have not undergone external peer review. We welcome independent replication and critique. All methodologies, raw data, and analysis procedures are published openly to enable scrutiny and verification by the research community.
Leads SIGI's research programme into LLM citation behaviour and generative search patterns. Responsible for research design, data analysis, and publication strategy.
Oversees SIGI's technical infrastructure, data collection systems, and multi-platform testing frameworks. Leads methodology development and research tooling.