Nimble, a New York City-based tech startup, just launched Web Search Agents, a retrieval system designed to make AI-driven research both more accurate and significantly cheaper. The new platform aims to solve a major bottleneck in enterprise AI: the high cost and inefficiency of forcing language models to sift through massive amounts of generic, irrelevant web data. (source)

The core change for your workflow involves shifting from generic search to domain-specific retrieval. Instead of a single model trying to understand everything, Nimble builds specialized retrieval models that learn the specific signals and standards of your unique business domain. This optimization allows agents to find the right information faster, which reduces the amount of multi-hop reasoning required and lowers your overall token usage.

The numbers behind the launch show a significant leap in efficiency compared to leading AI search alternatives: Read more: Prentis raises $100M for AI agents. Here is what to expect from your office workflow.

• 21% improvement in accurate web research.
• 51% reduction in token usage.
• 20x reduction in token costs reported by early customer Rox.