Oscar Täckström: From Enterprise Search to Intelligent Agents

The landscape of enterprise knowledge access is rapidly evolving from simple search and discovery, through intelligent assistants, and now towards reasoning agents. I will trace the progression of enterprise AI technologies along these three arcs focusing on the key drivers and give my perspective on where we’re headed and how we take this to heart at Sana.

We'll dive deep into the lessons we’ve learned from building Retrieval-Augmented Generation (RAG) in real-world contexts, examining how recent advances in test-time compute and agent architectures are reshaping our approach. Central to this exploration is how Large Language Models (LLMs) are fundamentally reimagining document understanding and information retrieval, enabling more nuanced, context-aware, and adaptive intelligent systems.

Oscar Täckström is the Chief Scientist at Sana, where he leads all AI efforts. His research background is in natural language understanding and machine learning with notable contributions to attention-based models for language and video generation. Before joining Sana, he co-lead the technical development at Inceptive, applying deep learning to mRNA molecule design. Prior to that he was a senior research scientist at Google. He obtained his PhD from Uppsala University in 2013 and his publications won best paper and test of time awards at NAACL 2013 and ACL 2023.

Sana is an AI company on a mission to advance how humans access knowledge with AI. Its products are trusted by leading enterprises like Merck, Hinge Health, Electrolux, and Svea Solar. Backed by top-tier investors including NEA, Menlo Ventures, and EQT Ventures, Sana has raised over $130 million to date.