A Product Market Fit Show | Startup Podcast for Founders
00:01:09 · Surojit Chatterjee, Pablo Srugo
[00:01:09] S0: Product market fit. Product market fit. I mean, the name of the show is product market fit. Do you [00:01:14] S1: think the product market fit show has product market fit? Because if you do, then there's something you just have to do. You have to take up your phone. You have to leave the show five stars. It lets us reach more founders, founders and it lets us get better guests. Thank you. [00:01:28] S0: Surajit, welcome to the show, man. Excited to be here. [00:01:31] S1: See, you've got a pretty crazy resume, VP at Google, you're a chief product officer at Coinbase, EIR at a sixteen z, and now you've started Emma, which is in the AI agent for enterprise world. We're gonna get into all that. We're gonna get into why you started it, how you started it, how you got there. But because this is the product market fit show, let me capture a moment in time. When would you say was the moment when you felt like you'd found true product market fit?
00:02:00 · Surojit Chatterjee, Pablo Srugo
[00:02:00] S0: For us, look, we are an enterprise software company. Right? We are building AI employees for largest enterprises. Just to give a little bit of background, right, for a startup of our size, like starting out, you are trying to get to Fortune 500 and telling them you're gonna touch their most sensitive data. Like their employee data, their finance data, these are areas where we go really deep. That's a very difficult kind of proposition to start with. So for us, the breakthrough was really, you know, one of our clients, Hitachi, a very large conglomerate. And they were trying to automate HR operations for five of their companies and create a shared service across them. And they have around 55,000 employees across these companies. So it's a very complex setup. You are touching sensitive data. You are on multiple cloud platforms at the same time. The agents need to coordinate and cooperate. For me, when we could deploy and showcase that it actually works, it was really the moment I felt that we have hit product market fit in terms of the scalability of the product, the governance, security, and the selling to other clients just became that much easier for us because it's credibility is everything in an enterprise context. [00:03:20] S1: And was this like when you signed the contract or once you've fully deployed and you were seeing the outcome?
00:03:20 · Pablo Srugo, Surojit Chatterjee
[00:03:20] S1: And was this like when you signed the contract or once you've fully deployed and you were seeing the outcome? [00:03:26] S0: When we've deployed and started seeing the outcome. I mean, signing the contract is only the first step in many of the enterprise deployments. What's happening is a lot of enterprises are doing a lot of POCs and pilots. So this was not a pilot. We already did a pilot with POC, we went through the whole process. Actual full deployments, touching production level data. And by the way, now this is not our largest deployment at all. Now we have other clients that are even larger and in production. But that was, in my head, was really the beginning of the journey. [00:03:59] S1: And maybe as a way, this could be like a perfect example to use to really understand what it is that you do. Like, what did they wanna accomplish with Emma, with the product? How did it all work as a way for us to really get what it is
00:04:09 · Surojit Chatterjee
[00:04:09] S0: that you do? So look, we are an AI employee company. So we build AI employees. We have a platform that allows us to build AI employees. What are AI employees? AI employees are basically teams of agents. Now a lot of people are seeing this in the real world and from a lot of other companies as well. We started working on it probably before anybody else. You know, two and a half years back, right out of the gate, we had the same vision. What do these AI employees do? They basically connect to multiple systems. So they have a think of a mesh of agents, like a team of agents. Each agent is skilled in in certain thing, and they are all coordinated using our own orchestrator and control plane. These agents will touch or operate different tools in the enterprise. It could be your Workday or Greenhouse or ServiceNow or Salesforce. We'll also connect to all kinds of data sources. The data doesn't need to be cleaned. It can be anywhere, SharePoint or Google Drive, database, wherever it is, data warehouse. These agents will understand the data, understand the policies, understand your process, reformat your process, kind of reimagine what the process needs to be, create an kind of an agentic workflow, and automate what I call the human glue between systems today. Every enterprise today has hundreds of applications, SaaS applications, and thousands of people who are acting as glue between those applications, moving data from one, pushing data somewhere else, creating a report, sending an email. Think about what people do day to day in an enterprise. So we are creating that intelligence and action layer across these applications and all this data, which basically frees up humans to do more value added work rather than doing the same similar mundane work every day.
00:05:59 · Pablo Srugo, Surojit Chatterjee
[00:05:59] S1: So maybe I mean, that's high level. Right? But, like, just to get to the specifics, like, what exactly did they want AI agents to do? [00:06:06] S0: Look, simple things to very complex things. Right? For simple things will be, hey, I am an employee in a in a country. Right? So let's say, I'll take give example of another client. Right? They have quarter of a million employees, 65 different country. Large services company, professional services company, Wipro actually. Their employees are very mobile because they do projects with customers and and so on. Very complex HR ops because employees are asking questions about, hey, what's my vacation policy? Oh, I am I was hired in Romania and now I am in Germany. What does my benefits look like? Right? Can I go to a doctor? Usually, of those questions will go to HR and they will answer. So that's questions are only one part of it, which is in itself is complex because there are, like, you know, 65 different level laws and policies and so on. Right? [00:06:59] S1: But that's more of a while getting the data's complex, that's more of like a chatbot type.
00:07:03 · Surojit Chatterjee
[00:07:03] S0: The chatbot. And there are actions. Somebody says, hey, I need to move to another location. That kicks off a whole workflow inside HR. Right? Which is like, okay, arranging their move, changing their tax form, maybe changing their comp, changing their management. There is a whole workflow orchestration that HR has to do. And we are working with another large company today. It's crazy. They have 65,000 job changes every month Wow. Within the company. Like, people either getting promoted, moving to another role, whatever it is. Right? Moving to another location, all kinds of things. These are massive global companies. And for each one of those 65,000, you imagine, there will be some humans help involved helping and orchestrating that change. Right? Can you automate that? And that automation requires it's not just a chatbot. Right? Multi step planning, orchestration, touching different systems of records, and helping the employee through that change. So anything from like, you know, book my travel. You know, a lot of companies, there's some team internally helping book travels. Right? Or time sheet management. So consulting companies in particular, they have to manage their time sheet very critically because that's how they charge their customers. Or book a cab even. Right? Because some of these companies are running large like operations team that are doing overnight shifts and then company provides some benefits. So you can, you know, take a company cab and go home or whatever. Right? So everything and anything all through a simple interface. And imagine that you don't need to go through any human or any tools. The biggest challenge today is proliferation of tools. So there are 100 plus tools in this company doing all this for doing all these different things. And there are thousand plus humans that you can reach out to. You also don't know who to reach out to sometimes. You're like, oh, my laptop is not working. See, an employee ultimately, they don't know or should is this falling into HR or IT or finance? My laptop is not working. My I want to change my payroll election. Right? Selections, you know, w two selection, whatever. Right? Or by, you know, deductions or under change by benefits.