The competition in enterprise AI is intensifying. Microsoft is embedding Copilot into Office, Google is integrating Gemini into Workspace, and companies like OpenAI and Anthropic are directly engaging with businesses. Moreover, nearly every SaaS provider is now offering an AI assistant.
Amidst this intense focus on user interfaces, Glean is pursuing a different strategy: establishing itself as the invisible intelligence layer that powers these systems.
Seven years ago, Glean began with the ambition to be the Google for enterprises, developing an AI-driven search tool capable of indexing and searching across a company’s diverse SaaS applications, from Slack and Jira to Google Drive and Salesforce. Today, the company’s focus has evolved from simply creating a superior enterprise chatbot to becoming the essential connective framework linking AI models with enterprise systems.
“The foundational layer we developed initially—a robust search product—required us to deeply understand people, their workflows, and their preferences,” Jain shared with TechCrunch during last week’s episode of Equity, which was recorded at Web Summit Qatar. “All this understanding is now crucial for developing high-quality AI agents.”
He highlighted that while large language models possess immense power, they are inherently generic.
Jain explained, “The AI models themselves lack specific knowledge about your business. They don’t know your personnel, your operational activities, or the products you develop. Therefore, it’s essential to integrate the models’ reasoning and generative capabilities with the unique context within your company.”
Glean’s core proposition is its ability to already map this essential business context, positioning itself as the intermediary between AI models and enterprise data.
While the Glean Assistant, a familiar chat interface powered by a blend of proprietary models (such as ChatGPT, Gemini, and Claude) and open-source alternatives, all grounded in a company’s internal data, often serves as the initial point of engagement for customers, Jain contends that the underlying infrastructure is what truly retains them.
Firstly, there’s model access. Instead of obliging companies to commit to a single large language model provider, Glean functions as an abstraction layer. This enables enterprises to seamlessly switch between or combine various models as their capabilities evolve. For this reason, Jain views OpenAI, Anthropic, and Google not as competitors, but as collaborative partners.
Jain stated, “Our product continuously improves by leveraging the innovations these companies bring to the market.”
Secondly, Glean provides robust connectors. It integrates deeply with essential systems like Slack, Jira, Salesforce, and Google Drive, allowing it to understand information flow and empower AI agents to perform actions directly within these tools.
And third, and arguably most critically, is governance.
Jain emphasized, “It’s imperative to establish a permissions-aware governance and retrieval layer that can deliver the correct information, while simultaneously identifying who is querying and filtering that information based on their specific access rights.”
For large organizations, this governance layer is crucial for transitioning AI solutions from pilot programs to full-scale deployment. Enterprises cannot simply feed all their internal data into a model and expect to resolve access and security issues later, according to Jain.
Equally vital is preventing AI models from generating incorrect or fabricated information (hallucinations). Jain explained that Glean’s system verifies model outputs against original source documents, provides line-by-line citations, and ensures that all responses adhere to existing access permissions.
The central question remains whether such a middle layer can persist as major platform providers intensify their efforts to integrate deeper into the technology stack. Microsoft and Google already command significant portions of the enterprise workflow landscape and are actively seeking to expand their influence. If their offerings, Copilot or Gemini, can access the same internal systems with identical permissions, then what role does a dedicated intelligence layer continue to play?
Jain contends that enterprises prefer not to be tied to a single model or productivity suite. Instead, they would opt for a neutral infrastructure layer over a vertically integrated assistant.
Investors have embraced this perspective. Glean secured a $150 million Series F funding round in June 2025, nearly doubling its valuation to $7.2 billion. Unlike cutting-edge AI labs, Glean does not require extensive compute resources.
“We operate a robust, rapidly expanding business,” Jain affirmed.