Large language models

Digitalis offers clients a comprehensive audit of their chatbot profiles including an assessment of the sources from which they are comprised. For simple keyword queries and more complex directional searches, we assess the accuracy of the content and we interrogate its sources to establish their relationship with your search engine results.

This process informs an actionable plan to ensure your Large Language Model (LLM) profile is contemporaneous and accurate, and does not convey any sensitive, negative or private information.

ChatGPT already attracts more than 300 million weekly users and Google expects 500 million weekly Gemini users by the end of 2025. Alongside traditional search engines – often used for distinct purposes – the most popular LLMs now constitute an increasingly significant global information resource and influential media channel for desk researchers.

Desk researchers are now accustomed as part of their workflow to soliciting chatbots for synthesised responses. LLMs are also used to inform niche research platforms. With their output interpreted as factual, understanding what these inherently imperfect answer-engines convey about your organisation is now critically important.

In 2023 Digitalis pioneered new research examining from which open-source information the most popular chatbots derive their information, including from your surface and deep web footprint, and how they formulate your LLM profile. We have since built a tool to automate this research and support us in gaining up-to-date insights into the continually changing prioritisation of sources cited by chatbots.

Despite powerful processing capabilities and the factual tone of their answers, many chatbot profiles contain inaccurate content. With their increasing adoption, leaving this aspect of your online reputation to chance isn’t an option.