Two years ago, Arvin Jain had to convince enterprises that his company's AI-powered search product was worth buying. Today, every major software company in the world is racing to sell something that looks a lot like what Glean already built.
Glean started in 2019 as a straightforward pitch: Google for your work life. Jain, who spent years as one of Google's early search engineers before co-founding data security company Rubric, wanted to make it easy for employees to find information scattered across dozens of internal tools. The company leaned on transformer models early and, almost by accident, became one of the first enterprise generative AI companies in the world.
The product has since evolved well beyond search. Glean now positions itself as an enterprise AI assistant and, more ambitiously, as a platform for building custom AI agents. A recent $150 million Series F round valued the company at $7.2 billion, up from $4.6 billion, reflecting how quickly the market has shifted from curiosity about enterprise AI to outright urgency.
The land grab replaced the sales pitch
Jain describes the shift in blunt terms. The company went from evangelising AI to operating in a total land grab. Enterprises are no longer asking whether AI can help their business; they are actively searching for the products that will define how AI gets deployed across their organisations.
That urgency has attracted serious competition. Microsoft with Copilot, Google with Gemini for Workspace, Salesforce with Einstein, plus Notion, Atlassian, and a growing list of others, are all fighting for the same territory. Jain says Glean competes daily with the 10 or 15 largest software companies in the world, all of which now consider themselves AI companies.
The middleware bet
Glean's strategic argument is that it does not need to beat those companies head-to-head. Instead, it wants to sit behind them.
The company's core pitch rests on three technology layers that Jain says any serious enterprise AI deployment requires. First, a model access layer that can draw on multiple foundation models, including Gemini, Claude, GPT, and open-source alternatives, rather than being locked to a single provider. Second, an integration layer that connects to the dozens of systems where a company's actual knowledge lives. Third, a permissions-aware governance layer that filters information based on who is asking, ensuring that sensitive data only reaches authorised eyes.
That governance layer is the piece Jain emphasises most. When an employee asks an AI assistant a question, the system needs to know not just where the answer lives, but whether that particular employee is allowed to see it. Glean's pitch is that seven years of building enterprise search gave it a deep understanding of how information flows inside organisations, and that understanding is now the foundation for building agents that can act on behalf of employees without creating security nightmares.
Powering Copilot from behind the scenes
The practical result is that Glean increasingly operates as middleware. A customer might use Microsoft Copilot as their front-end AI experience, but Glean sits behind it, pulling in data from Salesforce, internal wikis, and other systems to give Copilot the context it needs to produce a useful answer.
Some customers use Glean as a standalone chat product. Others use it purely as a data layer to power AI experiences built on platforms like Copilot Studio or Google's Vertex. Many do both, starting with the chat interface and gradually expanding into the platform capabilities as their AI strategy matures.
Jain frames this as a deliberate architectural choice. Rather than competing with Microsoft or Google on the user-facing layer, Glean wants to be the horizontal intelligence platform that makes every verticalized AI experience smarter. The productivity giants will bundle AI into their existing suites; Glean will supply the cross-platform knowledge and governance that those bundled tools cannot provide on their own.
Enterprises are already consolidating their AI stack
The early phase of enterprise AI adoption, where companies experimented with dozens of point solutions, is giving way to something more disciplined. Jain says enterprises are now thinking holistically about their AI architecture, recognising that onboarding too many tools creates the same dependency chaos they experienced with SaaS applications over the past decade.
The result is consolidation. Companies are narrowing their AI stack to a smaller core of platforms and defining which products will handle end-user experiences versus which will power the intelligence behind the scenes. Glean is positioning itself firmly in the latter category, arguing that a central horizontal platform is the only way to deliver AI safely, securely, and with proper governance at scale.
Agents are real, but humans still check the work
On the question of AI agents replacing workers, Jain is measured. Many companies are promising agent platforms that can replace employees in specific job functions, from customer service to HR operations. Jain says this is ahead of reality. Most agents still require human review because hallucination rates, even with the best models, remain too high for enterprises to accept fully autonomous operation.
Glean has built fact-checking technology that uses AI models to verify the output of other AI models, and provides features like hover-over source extracts so employees can quickly confirm that an AI-generated answer is grounded in real documents. In customer service, some Glean customers have achieved a 40% reduction in ticket resolution time using AI agents, but with humans still in the loop.
The more interesting shift, Jain argues, is happening at the leadership level. CEOs and executives are using AI tools to self-service information that previously required waiting for reports or briefings from their teams. The technology is not replacing executive assistants or staff; it is giving leaders direct access to project status, key risks, and operational details in real time, changing how decisions get made.
Reskilling over replacing
Jain pushes back on the narrative that AI adoption means mass layoffs. Companies are laying off employees in some areas, he acknowledges, but they are also hiring again as they work to right-size for new skill sets. The dominant trend he sees among large enterprises is a focus on reskilling their existing workforce to be AI-first, rather than replacing people with tools.
As for Glean's own trajectory, Jain says the company plans to stay private for at least the next couple of years. At $7.2 billion and growing, the IPO question is inevitable, but Jain insists the company is still early in its journey. Given how fast the enterprise AI market is moving, that framing might be the most telling detail of all.