A large language model (LLM) is a neural network trained on huge amounts of text, learning statistical patterns of human language in the process. Notable examples are Claude, GPT-4, Gemini, Llama and Mistral. LLMs can summarise, translate, answer questions, write code or extract structured information from unstructured sources.
In application and recruiting contexts, LLMs are used in many places: to parse CVs and job postings, to draft cover letters, to generate suitable interview questions, to summarise roles or to assess the "AI safety" of occupations. The combination of an LLM with well-curated data is often the deciding factor.
Risks matter: LLMs can produce factually wrong statements ("hallucinations"), reproduce biases from training data and breach data-protection requirements when sensitive data is fed into third-party models. In professional use, clear data-protection rules and reviews are indispensable.
Lunigi uses LLMs in the background to research and assess matching roles – the results appear compactly and transparently in the daily email digest.