Profile
The Question That Built a Company
Somewhere between his eleventh year at Google and his third at Coinbase, Surojit Chatterjee arrived at a question that most executives are too polite to ask out loud: if we've spent decades recruiting the most capable people, why do those same people spend most of their day on tasks a well-trained machine could handle? The answer - or rather the refusal to accept the non-answer - became Ema.
Ema isn't a chatbot. It isn't another dashboard. It's a universal AI employee - a system that integrates with the 200-plus tools already living in enterprise tech stacks and performs end-to-end workflows without a human in the loop. Customer support tickets. HR onboarding. Invoice processing. Prior authorization in healthcare. Chatterjee's thesis is that every complex, repeatable job a smart person currently does by clicking through five windows and two approval chains is, in principle, automatable.
"Why do we hire the smartest people and give them jobs that are so mundane?"- Surojit Chatterjee, Founder & CEO, Ema
The company emerged from stealth in March 2024 with $25 million in seed funding and 50 pre-built AI employees ready to deploy. By July, a $36 million Series A had landed - led by Accel and Section 32. By October, KPMG had made a strategic minority investment. The list of individual backers reads like a reunion of Silicon Valley's most consequential careers: Sheryl Sandberg, Dustin Moskovitz, Jerry Yang, David Baszucki.
EmaFusion blends 100+ public language models with domain-specific custom models. The result: an AI employee that picks the right model for each task - like a chef who can cook any cuisine on demand, not just the one dish the menu lists.
The Google Years - Building at Scale
Chatterjee spent 11 years at Google. That number understates the scope. He was among the founding members of Mobile Search Ads at a moment when "mobile" meant most people still doubted it would matter. It mattered. He then moved to Google Shopping, scaling it from an interesting side project into a multi-billion dollar revenue engine that now shapes how a significant fraction of global retail commerce gets discovered.
The patents followed - 40 of them, covering distributed computing, mobile advertising, location-based technologies, payments, machine learning, and enterprise software. A working engineer's body count, spread across a decade of genuine invention at one of history's most prolific technology companies.
What the patents cover
Distributed computing • Payments • Mobile advertising • Location-based technologies • Machine learning • Enterprise software - 40 US patents held, spanning the arc of his career from mobile ads to agentic AI.
Coinbase - Into the Chaos
Joining Coinbase in 2020 as Chief Product Officer meant signing up for one of the most turbulent IPO runs in modern tech history. Crypto went parabolic, went sideways, went everywhere - and in April 2021, Coinbase listed directly on NASDAQ. Chatterjee oversaw Product Management, Design, User Research, and Strategic Programs through the whole thing. The company's mission - make the cryptoeconomy accessible to millions - required building products fast under conditions that made "fast" feel like an understatement.
The stint gave him a close-up view of what it looks like when a company tries to scale faster than its operational infrastructure can follow. Thousands of employees. Thousands of workflows. Most of them stitched together with spreadsheets, email chains, and institutional memory that lived in people's heads. It's the kind of organizational friction that a product leader notices and can't unsee.
"We are at the cusp of the biggest change modern history has ever seen since the industrial revolution. If there was more automation in enterprises, the opportunity of wealth creation would be in the order of $1 trillion to $10 trillion."- Surojit Chatterjee, YourStory Interview, 2024
Ema's Architecture - The Chef Analogy
Chatterjee reaches for a food metaphor when he wants to explain what EmaFusion actually does. Imagine you have a Michelin-star chef in-house - one who can cook any cuisine you name. French on Monday, Italian on Tuesday, Indian on Wednesday. You don't hire a different chef for each meal. Ema works the same way: its proprietary model-blending technology selects from 100-plus public language models plus custom domain-specific models, routing each task to the most appropriate system rather than forcing everything through a single model's limitations.
Alongside EmaFusion sits the Generative Workflow Engine (GWE) - a patent-pending system for building enterprise workflows without writing code. An operations manager can describe a process in plain language and Ema constructs the automation. The 200-plus pre-built connectors handle the integration layer. Enterprise customers have reported 70-percent-plus ticket deflection rates in customer service deployments.
- EmaFusion technology blends 100+ AI models dynamically per task
- Generative Workflow Engine (GWE) - patent-pending, no-code workflow builder
- 50+ pre-built AI employees ready to deploy
- 200+ enterprise connectors for existing tool stacks
- On-premises deployment for regulated industries
- Microsoft Pegasus Program - invite-only enterprise incubator
The IIT Kharagpur Kid Who Went to MIT
Before any of this, there was Kharagpur. The Indian Institute of Technology campus in West Bengal - IIT Kharagpur - is among the most competitive undergraduate programs in India and one of the hardest-to-enter engineering schools anywhere. Chatterjee graduated with a B.Tech in Computer Science and Engineering. He then added an MS in Computer Science at SUNY Buffalo before an MBA at MIT Sloan's class of 2006.
In 2024, IIT Kharagpur named him a Distinguished Alumnus - an honor awarded to graduates who have made a demonstrable impact on their field. The timing was pointed: the same year Ema emerged from stealth and rewrote his public biography from "Coinbase ex-CPO" to "enterprise AI founder."
Responsible by Design
Chatterjee is explicit about one thing that distinguishes his framing from the more frictionless AI narratives in circulation. He uses the word "responsible" often. Not as a PR buffer, but as a design constraint.
"You can't develop AI blindly - it needs to be responsible. The software should be adaptable, learning from its environment and the enterprise it serves. Continuous evolution and adaptability are key for responsible AI."- Surojit Chatterjee, Unite.AI Interview Series
Ema's architecture reflects this: full visibility and governance, enterprise-grade security, on-premises deployment options, and systems that learn from human feedback rather than running unsupervised. The pitch to enterprise procurement teams isn't "trust the AI." It's "the AI earns trust, incrementally, with every task it handles correctly."
The Co-Founders and the Team
Chatterjee co-founded Ema with Souvik Sen - formerly VP of Engineering at Okta and a Google veteran in data, ML, privacy, and safety - who holds 37 US patents of his own. The founding team also includes Swati Trehan. Together, Chatterjee and Sen alone hold 77 patents. This is a company founded by people who have built things before, at serious scale, in serious environments.
The 130-person team operates from San Francisco and Mountain View, with operations in Bengaluru. Strategic partnerships span Microsoft, Wipro, Hitachi Digital Services, ISG, and NDI. The KPMG investment - rare for a global consultancy to make in an early-stage AI startup - signals that Ema has cleared the enterprise credibility threshold that most AI companies spend years trying to reach.