Research

Research perspectives

Our take on advances in generative AI, from model training and retrieval to reasoning and agents, with a focus on their value for companies.

  • Article · 2 min read

    Better search starts with the context of your documents

    A passage becomes easier to find when the search system understands where it belongs. Research on contextual embeddings explores how to preserve that information efficiently.

    Lukas BierlingDiscussing: Context is Gold to find the Gold Passage: Evaluating and Training Contextual Document Embeddings

  • Article · 3 min read

    Similar retrieval scores, very different costs

    A new benchmark compares ten LLMs with 26 embedding models across 37 tasks. The results show why companies should decide carefully which work a language model does.

    Lukas BierlingDiscussing: The Embedder's Dilemma: LLMs Are Better, but at What Cost?

  • Article · 2 min read

    Why business agents make a strong case for small language models

    Business workflows contain many recurring, well-defined tasks. Specialised models could make these steps cheaper to run and easier to control.

    Lukas BierlingDiscussing: Small Language Models are the Future of Agentic AI

  • Article · 2 min read

    Teaching small models to find the evidence that matters

    Finding a related document is only the beginning. Research on specialised rerankers shows how compact models can learn to judge whether a passage actually helps answer a question.

    Lukas BierlingDiscussing: Distillation and Refinement of Reasoning in Small Language Models for Document Re-ranking