I Built a $100M Company in 3 Years by Betting on AI Agents | Arvind Jain (Glean)
How AI agents will transform work, why every employee needs their own AI team, and the secrets behind Glean's rapid growth to $100M ARR
Dear subscribers,
Today, I want to share a new episode with Arvind Jain.
Arvind is the CEO of Glean and believes every employee should have a team of AI agents to help them get work done. We had a great chat about how AI agents will impact work and Arvind’s top lessons from scaling Glean to $100M ARR in 3 years.
Watch now on YouTube, Apple, and Spotify.
This episode is brought to you by Merge — Merge gives SaaS companies like Ramp and Drata a single API to launch over 200 product integrations fast. Book a meeting via www.merge.dev/peteryang and get a $50 Amazon gift card when you attend.
Arvind and I talked about:
(00:00) How to find job security with AI
(04:03) Everyone will manage a team of AI agents
(05:53) Is PM well positioned to thrive with AI?
(10:09) How anyone can build powerful AI agents now
(12:28) Three real barriers to enterprise AI adoption
(17:34) How Glean reached $100M ARR in only 3 years
(22:52) Making AI work with messy internal company data
(26:51) Critical skills you need to build to craft AI products
(33:33) Are 5-year plans completely worthless?
(36:10) Glean's vision is to give everyone an AI team
AI agents and the future of work
Welcome Arvind! You've had a long career at Google and startups. I’d like to start with a provocative question. In big tech, you have layers of management and very specific functions. Will AI fundamentally change this work environment?
AI will certainly change how everyone works, whether you're in a startup, a small company, or a large enterprise.
I think the work we do today — we'll be able to do 10 times as much in the future.
It's like using a calculator versus adding numbers manually. The same thing will happen with most of our work. AI will handle the manual tasks, but that doesn't mean we'll work less. History suggests we'll find new things to do, creating more advanced technologies and products.
In startups, roles tend to be broader. I've played many roles when starting companies — IT, HR, developer, salesperson, and even HR. There's less specialization in startups than in large companies.
AI will handle tasks across these functions. You can imagine a developer gaining a lot of mileage through AI to do things previously requiring other people.
So, will everyone become managers of AI agents?
Absolutely. That's a concept I'm excited about. Take a CEO with an amazing team — chief of staff, executive assistants, a capable executive team, and a coach. This helps allow a much bigger impact than as an individual contributor.
Or consider Roger Federer. He plays with a racket and ball just like I do but has a team of 50 people, ensuring he's well-practiced with the right coaching. He delivers world-class performance, and I don't.
When you bring this to AI, everybody will have a different experience.
We will all have this amazing team of agents around us who will handle most of our work because they understand who we are and help us get better.
That's the future we're looking at. You can leave school and have that luxury rather than wait years.
Many listeners are product managers handling internal alignment and documentation. Is this role well-positioned for the future? How do you get your PMs to upskill for this AI agent future?
Our company's product managers are very heavy users of Glean. For context, Glean is like a more enterprise or powerful version of ChatGPT inside your company. It helps with world knowledge through models and web search, but also with internal knowledge it's connected to.
We also have an agent platform for specific day-to-day processes. Product managers use more agents than any other function — more than engineers, salespeople, or support. Why? Because their job constantly involves working with information from different people, customers, and product usage to make data-driven decisions.
Take feature prioritization. Before, you'd decide priorities by working with important customers through 10-15 meetings to build a roadmap.
Now, our team has access to all customer conversations ever held — thousands of them.
They've built an agent that listens to all call recordings for requirements or feature requests and classifies them into different themes.
The agent creates a spreadsheet with all feedback, then summarizes, synthesizes, and identifies top themes. Previously, you worked with limited information from a few meetings. Now, they leverage collective organizational knowledge that was untapped before AI. This makes roadmap planning more confident.
That creates more thinking time versus just gathering information. Can these agents automatically summarize information daily?
Yes, absolutely. Building agents in Glean is trivial. You don't need to be a developer or product manager. You could be a salesperson describing tedious work in natural language: "This is my work. Can you do it for me from now on?"
We'll build that agent. It's already connected with your enterprise systems and data, so it can provide the right information, use LLMs, and take action to complete the work.
There are two types of agents. One you actively use and ask to do work for you. The other runs autonomously in the background, triggered by events. For example, when a new lead enters your marketing system, it enhances it with additional information.
So the second type works like Zapier but for internal company systems — intelligent automated workflows?
Yes, think of Zapier with their extensive actions library. But an agent is more than that. There's logic and thinking involved. When an agent works, it reads information from enterprise systems and takes action, but much of it is thinking, analysis, and synthesis. That's where AI shines. That's what's new in enterprise automation — automating things that needed humans before because they required intelligence.
Glean’s secret to reaching $100M ARR in 3 years
Let's talk a little about enterprise adoption of AI. Why do you think some companies have been slow to adopt AI? Is it security concerns or simply time?
Large enterprises take longer to adopt AI technologies. Even for us serving large enterprises, we must be careful about the technologies we use internally. If something bad happens, we risk our customers. I see a few key friction points:
Security. You need to respect governance requirements, security, and permissions architecture. Don't expose information — even internally — to employees who shouldn't have access. Enterprise data is often private.
Inertia. In large enterprises, business processes are solidified. When you're on a treadmill of tasks, you don't have time to think about doing something differently with AI.
Education. Many employees find AI inaccessible. An HR person asked to use AI or build an agent might say, "What are you talking about? I've never built anything like that."
Confusion. CIOs face thousands of startups and large companies claiming to be AI companies. Enterprises must think long-term while wondering which LLM providers will survive or which startups will exist in a year or two.
What has been Glean's secret sauce for enterprise adoption and hitting $100M ARR?
