As Microsoft weaves Copilot into Office and Google pushes Gemini across Workspace, the enterprise AI race is increasingly focused on who owns the user interface. Glean is taking a different approach. Rather than competing to be the primary assistant employees talk to, the company is positioning itself as the system that connects AI models to the internal data and workflows that actually power organizations.
In an interview on TechCrunch’s Equity podcast at Web Summit Qatar, CEO Arvind Jain described how Glean’s strategy has evolved beyond its original identity as an enterprise search tool. The company, founded seven years ago, initially set out to index and search across workplace software like Slack, Jira, Google Drive, and Salesforce. That early work mapping how information flows inside companies now serves as the backbone of its AI ambitions.
Building effective enterprise AI, Jain argues, requires far more than access to a powerful language model. “The AI models themselves don’t really understand anything about your business,” he said. They don’t know who works where, what teams are building, or how information is structured internally. To make AI genuinely useful at work, companies have to connect those models to rich organizational context.
Glean’s platform is designed to provide that connective layer. Its Glean Assistant offers a familiar chat interface powered by a mix of leading proprietary models — including ChatGPT, Gemini, and Claude — alongside open-source alternatives. But Jain says the chat window is only the entry point. The real value lies underneath, in the infrastructure that links models to enterprise systems while respecting permissions and governance rules.
One pillar of that infrastructure is model abstraction. Instead of locking customers into a single AI provider, Glean enables enterprises to switch between or combine models as capabilities change. Jain says he does not view OpenAI, Anthropic, or Google as direct competitors in this context. “Our product gets better because we’re able to leverage the innovation that they are making in the market,” he said.
Another is deep integration. Glean connects directly to tools like Slack, Jira, Salesforce, and Google Drive, mapping how documents, messages, and workflows intersect. That allows AI agents not only to retrieve information but also to operate inside those systems.
Governance is central to the pitch. Large organizations cannot simply load all their internal data into a model and worry about permissions later. Glean’s system builds a permissions-aware retrieval layer that tailors responses based on who is asking the question. It also verifies outputs against source documents and generates citations to reduce hallucinations and ensure responses align with existing access rights.
In highly regulated or complex environments, that governance layer can determine whether AI tools remain pilot projects or become company-wide infrastructure.
The open question is whether this neutral intelligence layer can hold its ground as platform giants move deeper into enterprise AI. Microsoft and Google already control large portions of the workflow surface area. If Copilot or Gemini can reach into the same systems with comparable access and controls, enterprises may question the need for an independent intermediary.
Jain contends that many companies want flexibility. Rather than tying their future to a single model or productivity suite, he argues enterprises prefer infrastructure that preserves optionality as AI capabilities evolve.
Investors appear to agree. In June 2025, Glean raised a $150 million Series F, nearly doubling its valuation to $7.2 billion. Unlike foundation model labs, the company doesn’t need to spend heavily on training models. Its focus is orchestration — connecting models, data, and governance into a cohesive enterprise system.
As the enterprise AI market matures, the competition may shift from who builds the smartest assistant to who controls the layer that makes those assistants actually work inside real organizations. Glean is betting that context — not just model capability — will define that next phase.
This analysis is based on reporting from TechCrunch.
Image courtesy of Glean.
This article was generated with AI assistance and reviewed for accuracy and quality.
