As competition among frontier model companies intensifies and open-weight models such as Kimi K3 and Qwen3.8-Max continue to grow more capable, model routing has become a critical part of AI deployment. We recently saw Stripe acquire OpenRouter for more than $7 billion, but this trend is just as strong in the enterprise sector.
Glean was co-founded and is led by Arvind Jain, a former Google Distinguished Engineer, and focuses on helping large organizations adopt AI. Last June, following its $150 million Series F funding round, Glean reached a valuation of $7.2 billion. This year, the company’s annual recurring revenue (ARR) reached $300 million, tripling in just 15 months.
One of Glean’s missions is to choose the right model for each task—and even determine whether a large language model (LLM) is actually needed in the first place.
“One important goal for Glean is to avoid using LLMs for tasks that don’t require them,” Jain told Latent Space. “Sometimes you’ll see someone enter a query in Glean just to add or multiply two numbers. They could simply use a calculator.”
What Glean primarily wants to do, however, is provide enterprise employees with what Jain calls ‘a truly powerful personal collaborator.’ That means Glean needs to become a meta-level agent framework—a meta-harness—built on top of leading large language models.
