“Every worker will have a team of assistants,” says Glean CEO Arvind Jain

“The future that I see with AI is that every person who works is going to have this amazing team of assistants, co-workers, and coaches around them that will make them a lot more effective,” said Arvind Jain, CEO of Glean, during a recent conversation on the Matt Podcast.

At Glean, that vision is not just aspirational. It is being operationalized inside companies already. Founded in 2019, Glean began by addressing a fundamental challenge Jain had witnessed throughout his career: despite the proliferation of digital workplace applications, employees increasingly struggle to find the information they need to do their jobs.

“Finding information at work has become increasingly difficult,” Jain said, reflecting on his time at Rubrik and Google, where even advanced technology environments failed to connect employees seamlessly with knowledge.

Glean’s first product, a secure and permission-aware enterprise search engine, focused on integrating data across hundreds of workplace systems like Salesforce, Confluence, Jira, and Workday. From the beginning, the company emphasized tight security surfacing only information employees were authorized to access while prioritizing relevance based on document freshness, subject-matter authority, and engagement signals. Language models supported retrieval but were used within strictly governed enterprise frameworks.

By 2022, as large language models matured, Glean expanded its platform from retrieval to direct answer generation. Glean Assistant enabled employees to ask natural language questions and receive synthesized responses grounded in internal enterprise data. It combined the company’s deep integrations with retrieval-augmented generation techniques to deliver more than just search results, delivering answers.

Yet even as the excitement around AI capabilities surged, Jain remained pragmatic about its limits. The narrative of fully autonomous AI agents replacing humans did not align with what he saw happening inside real organizations. “Agents are still better run in a supervised manner where a human is in charge and looking at the work of the agent,” he said.

Against this backdrop, Glean launched Glean Agents in February 2025, a significant extension of its product vision. Glean Agents provide a framework for building, deploying, and governing AI agents that automate business processes across HR, sales, engineering, legal, and more. The platform allows employees, not just developers, to create task-specific agents using simple natural language prompts, with deeper customization available for more complex workflows.

“Those closest to understanding specific business processes are best fit to capture and automate them,” Jain said, describing the philosophy behind the agent framework.

Glean Agents operate through a structured tool-usage model. Agents retrieve live data from enterprise systems, reason over it using large language models, and take appropriate actions, such as updating CRM entries, logging tickets in HR platforms, generating client communications, or analyzing operational trends. Structured data analysis is integrated directly into the platform, enabling users to query live databases such as Salesforce, Databricks, or Jira without manually coding integrations for each task.

The platform’s architecture connects structured and unstructured enterprise data with external web information and LLM knowledge bases, creating a unified base for agents to reason and act securely. Universal knowledge access ensures that agents can tap into the breadth of enterprise data without breaching governance policies.

Security remains foundational. Glean’s active governance layer continuously scans over 100 integrated applications for sensitive oversharing, automatically remediating potential compliance risks. Rather than treating AI as an external layer that ignores enterprise policy, Glean embeds security and permissions management into every agent action.

In practice, companies are already deploying Glean Agents for critical workflows. Zillow built a career growth analysis agent to help employees understand promotion pathways by interpreting performance evaluations. Miro uses agents to personalize sales outreach messaging, reducing the time spent crafting individualized emails by 80 percent. Deutsche Telekom deployed a concierge agent to automate IT and HR service interactions for more than 80,000 employees.

Within Glean’s own teams, agents have become integral to operations. Product managers analyze customer call transcripts at scale to guide development decisions. Engineers automate pull request review documentation. Sales teams prepare for meetings using agents that aggregate client and industry research. In each case, AI enhances human workflows rather than replacing human judgment.

Jain also addressed the broader AI model landscape during the conversation. Enterprises today mostly rely on closed models like OpenAI’s GPT, Anthropic’s Claude, or Google’s Gemini, particularly during the early stages of AI adoption when accuracy and reliability are paramount. Over time, cost pressures, data control needs, and customization goals are pushing organizations to consider open-source models.

Glean remains LLM-agnostic, supporting multiple model providers while routing tasks dynamically based on complexity, speed, and cost considerations. Recent integrations include Gemini 1.5 Flash and other frontier models. Jain described Glean’s operational principle in an environment where AI technology evolves constantly: “The technology environment is unstable,” he said. “You need to be able to assess new models within minutes.”

Rather than being forced into single-model dependencies, Glean’s system flexibly adapts, maintaining consistent user experiences even as underlying models change.

Glean’s expansion from search to Assistant to Agents has been methodical. The platform is designed horizontally, allowing enterprises to scale from a few agents to thousands without creating a fragmented sprawl of disconnected AI systems. Centralized governance, unified permissioning, and cross-functional reach allow Glean to act as a single AI infrastructure layer across an entire organization.

While much of the broader AI industry focuses on hypothetical transformation, Jain keeps Glean’s philosophy grounded in practical outcomes. When asked about how the company thinks about competitive moats, he responded directly: “You just work hard on solving user problems.”

Today, Glean Agents have powered over 50 million automated actions across customer organizations. The company’s focus remains consistent: helping employees work faster, more effectively, and more intelligently, not by removing human oversight, but by strengthening it with better tools.

“We want to be that team of AI agents around every individual that helps you do your great work,” Jain said.