May 15, 2025 · 31m · no-priors
No Priors Ep. 115 | With Glean Founder and CEO Arvind Jain
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of No Priors, Glean CEO Arvind Jain joins hosts Elad Gil and Sarah Guo to discuss how transformer architectures, cloud SaaS connectivity, and information retrieval have revitalized enterprise search. Jain details Glean's evolution from a knowledge index into an autonomous workflow agent platform while sharing actionable insights on data governance, category creation, and founder conviction.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 19% of the talking time here. How this is scored →
The hosts as informed peer 4.5 Guest teaching 3.8 Guest disagreement 1.1 The hosts pushing back 0.4
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 8:58 Arvind counters infinite context window assumptions
Arvind politely but firmly rejects the prevailing Silicon Valley idea that infinite context windows will render information retrieval and search ranking unnecessary.
Hardest push from the hosts ▶ 8:42 Sarah questions the long-term need for bespoke search infraSarah directly questions whether traditional search signals like freshness and authority will remain relevant or just get subsumed entirely by base foundation models.
Biggest teaching moment ▶ 3:25 Arvind explains vector embeddings are not enoughArvind details how vector similarity alone fails in legacy corporate environments without document recency, authority signals, and enterprise metadata.
The host holds their own ▶ 27:11 Elad synthesizes Glean's structural advantageElad displays deep architectural expertise by crisply synthesizing the three foundational shifts—internal IT need, SaaS API maturity, and transformer embedding breakthroughs.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| The Paradigm Shift from Keyword Search to Transformers | 4 | 4 | 1 | 0 | Sarah and Elad frame the evolution of search around LLMs and transformers. Arvind explains how early Glean was to BERT embeddings and points out that semantic vector search alone is insufficient without authority and freshness signals. | |
| How SaaS and Cloud Scalability Solved Enterprise Search | 6 | 4 | 1 | 1 | Elad and Sarah demonstrate domain familiarity by referencing failed historical attempts like FAST and hardware solutions like the Google Search Appliance. Arvind elaborates on how modern SaaS APIs and cloud scale enabled a turnkey product. | |
| Why Information Retrieval Outweighs Brute-Force Context Windows | 5 | 5 | 2 | 1 | Sarah probes whether infinite context windows will make traditional IR pipelines obsolete. Arvind pushes back against the brute-force context window hype, explaining using human reasoning analogies why structured, curated retrieval remains essential. | |
| Navigating Enterprise Access Control, Permissions, and Governance | 4 | 4 | 1 | 0 | Sarah raises questions around fine-grained permissions and end-user adoption patterns in enterprise search. Arvind explains the necessity of enforcing identity-aware access control so models do not leak private corporate data. | |
| Category Creation Hurdles and the Unexpected Fear of Search | 4 | 4 | 1 | 1 | Elad asks about unexpected hurdles transitioning from Rubrik to Glean, and suggests LLMs could classify sensitive data. Arvind describes the unexpected paradox where companies feared effective search because it exposed broken internal permissions. | |
| Go-To-Market Dynamics: Direct Enterprise Sales Versus PLG | 4 | 3 | 1 | 0 | Elad contrasts bottom-up product-led growth with top-down enterprise sales motions. Arvind explains why enterprise search inherently requires whole-company indexing rather than single-seat viral adoption. | |
| Ignoring Market Skepticism and Relying on Founder Conviction | 4 | 3 | 1 | 0 | Sarah asks how founders should evaluate negative market priors. Arvind candidly advises that over-analyzing prior failures leads to analysis paralysis, advocating conviction when real customer pain persists. | |
| The Long-Term Vision for Enterprise AI and Personal Co-Workers | 5 | 3 | 1 | 0 | Elad synthesizes the three pillars powering Glean's moat and asks about adjacent opportunities. Arvind outlines his long-term roadmap centered on AI coworkers and coaches for every employee. |
Statements from this episode (20)
Glean Used Transformers for Semantic Matching in Version One
“The version, one of our product actually already used transformers for semantic, you know, matching”
Glean Built Custom Enterprise Embeddings on Top of BERT
“We started with this BERT model that Google had put in open domain, which was trained on all of the internet's, you know, data and knowledge. And we would then take those models and then for every customer of ours, we'd actually build custom embeddings, you kn…”
Jain: Vector Search Alone Cannot Solve Enterprise Search
“There's a lot of focus on embeddings and vector search over the last few years, but that's actually only one part of, ah, building a good search system. Because if you think about an enterprise, ah, imagine a company that has been around for a few decades. You…”
Jain: Pre-SaaS enterprise search failed because turnkey products were impossible
“Most of the companies started in the pre-SaaS world, they failed because you could just couldn't build a turnkey product. But SaaS actually allowed you to actually build something, you know which is my insight.”
Jain: One of Glean's largest customers has over 1B documents
“One of our largest customers. They have more than one billion documents inside their company.”
Jain: The entire internet was one billion documents in 2004
“In 2004, the entire internet was actually one billion documents.”
Jain: Enterprise search requires transformer models due to scarce user signals
“On the web, even if you don't have good semantic understanding, there is so much that you're going to learn from people's behavior because, you know, you have a billion people, you know, coming and using your product. In the enterprise, you don't have that lux…”
Glean CEO: AI models are nowhere near replacing retrieval via infinite context
“I mean, I think there's always this thought of that, like, you know, the models will have near infinite context windows and you can just give them everything and they can figure things out automatically. But I don't think, you know, like they're anywhere close…”
Jain: Models reason significantly better when data is pre-organized
“There is a good amount of, ah, work that you have to do and present the information, ah, to the model in some, you know, in some organized fashion. That's when they're going to actually do a much better job reading that information, reasoning over it, and givi…”
Jain: Enterprises Care More About Process Automation Than Personal AI Assistants
“Businesses are actually a lot more, more interested in, not in that, but in actually thinking about how they can transform their company with AI, how they can take specific business processes, you know, where they're spending a lot of money and how, how do the…”
Jain: About 90% of enterprise knowledge is private within a company
“Most of the knowledge, I should say, like, 90% of the knowledge inside the company is private in some shape or form inside, within your company.”
Jain: Users adopted Glean search immediately but struggled with conversational Assistant
“Everybody has been trained over the last 20 years to actually type in, you know, one or two keywords, like Google has sort of taught us, you know, on what search can do. So with search, we never had a problem. Like, you know, we launched our product, we had, l…”
Jain: Glean faced nonexistent enterprise search budgets, unlike Rubrik
“In Rupert, we're an established market, like the budget, several dollars, and you had to actually replace an old technology with a new technology. Here we were in a market where we had no budgets. There was no concept of buying a search product in the enterpri…”
Jain: Companies feared effective search because it exposed internal governance gaps
“We started to hear from businesses that, oh, I'm scared of good search. I don't want a good search product in my company because I have all these governance gaps. I have like, you know, sensitive information all over the place.”
Jain: Enterprise search requires top-down sales because full-corpus indexing is expensive
“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 your…”
Jain: Startups should run PLG and enterprise sales simultaneously
“So the right recipe for me, like, you know, if I had a choice, I would actually start both the motions simultaneously. Like, I don't want to actually Say that, look, you know, for the first three years, I'm, I will actually focus, you know on just being PLG an…”
Jain: Looking closely at market priors leads founders to kill their ideas
“The more you look at priors, the more you're going to actually, likely you're going to actually ultimately kill your own idea.”
Jain: Finding internal information inside Google was super hard
“Even at Google, like, it was a big joke, you know, always we had internally, like, you know, All of us were spending all of our time making it easy for people to find things, but not us internally at Google. It's super hard to find anything inside the company.”
Jain: Missing enterprise knowledge is a bigger problem than AI hallucinations
“People talk about hallucinations as a big problem with AI models. You know, we feel like, you know, a bigger problem for us is not even hallucinations. It's about Like, you know, most of the times you can't even, you know, find that information. Sometimes it's…”
Jain: Workers will have personal AI teams performing 90% of their work
“And one thing that's going to fundamentally happen is that each one of us is going to have this amazing team Of, you know, call it assistants, coworkers, coaches that are truly personal to you. And, you know, you're always surrounded by that team. And this tea…”