Welcome to Tech Beat, your daily pulse on the stories shaping the digital world.
In the enterprise AI space, a conversation at HumanX is drawing attention. Neo4j's CTO Philip Rathle made the case that AI agents relying solely on trained models are a poor fit for business environments, where data goes stale fast. His proposed fix, Graph RAG, pairs vector search with knowledge graphs to keep agents accurate and contextually grounded.
On a different kind of intelligence, a startup called Thinking Machines is rethinking how AI conversations actually work. Right now every model operates in turns — you speak, it responds. Thinking Machines wants to build a system that processes your input and generates a response simultaneously, making interactions feel less like texting and more like a real phone call.
And in networking news, a veteran architect named James Thain has submitted a draft proposal to the IETF for something he's calling Internet Protocol Version Eight. Rather than replacing IPv six, Thain wants to improve on IPv four's foundations directly. He's hoping to crowdfund a testbed to prove the concept, though the proposal currently carries no official standing.
Those are the stories moving the needle today. Keep surfing. Tech Beat out.
