AI is Making Enterprise Search Relevant, with Arvind Jain of Glean
Summary
In this episode of No Priors, Arvind Jain, founder and CEO of Glean, discusses how large language models (LLMs) and AI are revolutionizing enterprise search by moving beyond traditional keyword-based methods to semantic, conversational assistants tailored for internal company data. He explains that while foundation models such as GPT-4 provide a strong starting point, substantial customization, including instruction tuning and integration with enterprise-specific knowledge, is essential for delivering relevant and actionable results in secure environments. Unlike public search engines, enterprise AI search must rigorously enforce permissions and governance to prevent sensitive data leakage, making security a core product feature. Glean’s approach transforms enterprise search into a personal AI assistant embedded within employees’ daily workflows, helping automate specific business functions rather than just answering Q&A. However, adoption is challenged by users’ entrenched habits with keyword search, requiring incremental education and motivation to leverage AI capabilities effectively. Jain also reflects on the unique challenges of building Glean compared to his previous ventures, including market immaturity, absence of budget lines for enterprise search products, and pervasive customer fears related to governance gaps exposed by powerful search tools. These obstacles necessitated not just innovative technology but evangelism and market creation to educate buyers and build demand. He further elaborates on the difficulties in scaling such a company-wide product that requires indexing substantial internal data and the consequent limitations of a pure product-led growth model, advocating instead for a hybrid go-to-market strategy combining PLG with direct enterprise sales. The episode concludes with reflections on leadership growth transitioning from engineering to CEO and the future potential for AI assistants fundamentally changing workplace productivity and business operations.
Key Takeaways
- 1Foundation LLMs like GPT-4 serve as powerful bases for enterprise search but require significant customization through instruction tuning and integration with internal enterprise data to deliver effective, relevant AI search and assistance.
- 2Enterprise AI search is evolving from traditional, brittle keyword-based search engines to AI-powered, ChatGPT-like personal assistants designed to seamlessly weave internal company knowledge into employees’ workflows as proactive 'sidekicks'.
- 3Successful enterprise AI implementations focus on transforming specific internal business processes through consequential, functionally embedded AI applications rather than offering generic Q&A capabilities.
- 4Handling data security, access control, and governance is a paramount, complex challenge in enterprise AI search, necessitating permission-aware architectures that prevent unauthorized access and information leakage.
- 5Users face adoption challenges in enterprise AI due to legacy habits of short keyword queries, requiring incremental exposure, role-specific prompts, and education to enable effective natural language interaction and maximize value.
- 6The enterprise AI market demands demonstrable ROI and efficiency gains, requiring AI product developers and vendors to focus on measurable business outcomes rather than exploratory features, balancing innovation with strong governance and user education.
- 7Launching Glean's enterprise search product involved substantial market creation efforts due to the absence of established budgets and familiar purchasing behaviors for such solutions, necessitating evangelism and education to build demand.
- 8Fear of exposing sensitive enterprise information through powerful search capabilities paradoxically creates resistance to adoption, prompting Glean to evolve into a security-conscious solution embedding document classification, sensitivity flags, and permission enforcement.
- 9The complex and company-wide nature of enterprise search products like Glean necessitates broad organizational deployment and precludes limited, individual user licensing, complicating product scalability and influencing go-to-market strategies.
- 10A hybrid go-to-market strategy combining product-led growth with direct enterprise sales is optimal for AI-powered enterprise search, balancing scalability ambitions with the realities of complex deployment, customization, and organizational buy-in.
Notable Quotes
""You can think of the Glean Assistant as a personal assistant that you're actually giving to every employee in your company. It's your sidekick, always available to help you with whatever questions or tasks you have, using all of your company's context and data to assist you with your work. But businesses are much more interested in how they can transform their specific business processes with AI, not just answering questions. They want to automate and improve workflows.""
""With enterprise information, 90% of the knowledge inside the company is private in some form. That means you cannot simply build a model inside your enterprise by dumping all of your internal company's data and knowledge into it and making that model available to everybody. Because if you do that, you're leaking information. Access controls and permissions are fundamental to how this will work.""
""AI is actually very unintuitive. For most people, you have to expose these capabilities incrementally. People have been trained over the last 20 years to type in one or two keywords, like Google taught us. But with assistants, people don't know what to do with it initially. You have to motivate and educate users to use AI in their daily work.""
""Businesses are excited about AI and have a lot of dollars to spend. But they also ask for ROI — what are the efficiency gains, top-line improvements, and returns? Any AI experience has to think about security, governance, and permissions at a fundamental level to hold out AI safely inside your enterprise. Education is often overlooked but is critical for success.""
"Like in Rubrik we’re an established market, like there were budgets, there were dollars and you had to actually replace an old technology with a new technology. Here, we were in a market where, uh, we had no budgets. There was no concept of buying a search product in the enterprise. And everybody thought that, yeah, like this is an important problem, but I'd like, you know, it's not a line item in my business priorities. You know, it's a vitamin, it's a painkiller. People are living without it."
"People found like salaries of other people, you know, those, like in one of our customers, somebody found a sensitive M&A doc that was, you know, or something that was, you know, not yet happened. And you start like, so people like you actually were very, very scared of actually having good search."
"We actually were forced to build that. We were forced to actually go above and beyond respecting permissions in individual systems to knowing who you are, what you're asking. Like, you should have the right to even ask the question or like, you know, when the information comes back, like, does it feel, you know, safe enough for us to show it to you? So we actually, in fact, like in that sense, you know, we actually ended up becoming a security product."
"I didn't actually, I wasn't the CEO. I ran R&D as one of the founders of the company. And here I had to actually learn how to become a CEO. And I don't think I've learned it yet. And like, you know, that's a constant, you know, challenge and like, you know, learning that I go through because fundamentally, like, I'm still an engineer. Everything I do, like, you know, like, that's the mindset that I have."
"But the problem is like, you know, with our product, it is by definition a company-wide product. Like it's not like, you know, we cannot offer the product to one individual inside a company. Even one person, you know, their search needs require us to actually search over all the entire company's information for them. So it's expensive. You have to actually index, you know, all of the company's data and knowledge."
"I'm an engineer and I wanted the company to have engineers and then product should sell itself, you know, on the web. Who doesn't want that? It was something that, you know, was a desire for us. But the problem is like, you know, with our product, it is by definition a company-wide product."
"So the right recipe for me, like, you know, if I had a choice, I would actually start both the motions simultaneously. Like, I won't actually say that, look, you know, for the first three years I'm, I would actually focus, you know, on just being PLG and then bring enterprise sales later because you're actually leaving a lot, you're leaving a lot on the table. Timing, timing matters always. And so you have to sort of like start the motions at the same time."
"I think as engineers, like, you know, there are, first of all, there are always doubts. Like, you know, the more you look at priors, the more you're going to actually, likely you're going to actually ultimately kill your own idea. There is a lot, like, you know, sometimes. Everything's been tried."
"A lot of things have failed and I think there are, like, for any given idea, like, you know, there are 10 reasons why it won't work. Like, as you start to go into details. Sometimes, like, a more simpler approach is helpful, you know, which is, well, there's a problem. Like, you know, you talk to people, they have and they feel this pain, and which clearly means that nobody is actually yet solving, you know, that because the pain exists."
"You know, I'm an engineer by training myself and I'm naturally trained to question and like, there's a lot of self-doubt in my mind. So, I don't know what happened to me when we started Glean, because, you know, there were all these people saying, like, no, not do it. And somehow they couldn't discourage me. Like, you know, I just felt that this was an exciting problem."
"At Google, it was a big joke, you know, always we had internally, like, all of us were spending all of our time making it easy for people to find things, but not us internally at Google. It was super hard to find anything inside the company."