Hane Labs is a Swedish software consultancy operated by Oskar Hane, a figure who is synonymous with the integration of graph technology and generative AI. While the broader market is currently saturated with horizontal agent platforms and general-purpose orchestration layers, Hane Labs occupies a distinct, high-leverage niche. It focuses on the architectural bridge between large language models and structured knowledge, specifically through the use of Neo4j and graph databases. Hane is a Senior Staff Software Engineer and the GenAI Tech Lead at Neo4j, and his private lab functions as a vehicle for applying these high-level technical insights to bespoke client projects.

The shift to GraphRAG

The foundational thesis of Hane Labs is that vector databases are an incomplete solution for the next generation of AI agents. Most current agent implementations rely on standard Retrieval-Augmented Generation (RAG), which uses vector embeddings to find relevant text. However, Hane’s work identifies the limitations of this flat approach. He advocates for GraphRAG, a methodology where agents interact with a knowledge graph to understand the intricate relationships between entities. This structure provides a level of determinism and traceability that vector-only systems cannot match, and it is essential for agents tasked with complex reasoning or data-heavy decision-making.

Technical pedigree and implementation

Hane’s career provides the necessary weight to this approach. His long tenure at Neo4j has given him a front-row seat to the evolution of data storage and retrieval. Before the AI surge, he was already building high-scale systems; now, he is one of the primary architects showing the world how to use graphs to ground LLMs. This experience allows Hane Labs to operate not just as a coding shop, but as an architectural consultancy. He helps organizations transition from simple chat interfaces—which often suffer from hallucinations and lack of context—to sophisticated agentic workflows that can query and reason over a verified web of enterprise facts.

The one-man lab model

Operating as a one-man show from Sweden, Hane Labs maintains a level of selectivity that is rare in the venture-backed startup world. This independence allows Hane to stay at the technical frontier, experimenting with how agents can best perform entity extraction and relationship mapping. This is a critical piece of the agent stack that remains unsolved for many: how to turn a pile of PDFs into a live, queryable knowledge base that an agent can actually use. Hane Labs provides the specialized knowledge required to build these memory layers.

Market position

In the competitive landscape, Hane Labs occupies the space between academic research and enterprise implementation. It does not compete with the model providers like OpenAI or the large-scale cloud providers. Instead, it stands apart from generalist AI consultancies by offering deep expertise in graph-native architectures. For the AI agent ecosystem, Hane Labs is a proof point that the future of agents is not just about the model, but about the structure of the data the model can access. By focusing on the knowledge part of knowledge work, Hane ensures that agents are grounded in reality rather than just statistical probability.