Participated · Equity · Mar 2024

Ema builds an agentic AI system—branded as a "universal employee"—that uses collections of AI agents called Ema’s Personas to perform complex, multi-step tasks autonomously. Its platform adapts to enterprise-specific data and workflows and can leverage multiple LLMs based on accuracy, cost, and performance needs. Ema emphasizes operating within company infrastructure to maintain data privacy and security while providing explainable and verifiable outputs. The product is designed to address common LLM shortcomings such as inaccurate outputs, data security risks, and a lack of domain-specific insight. Customers and partners include Moneyview, TrueLayer, and Envoy Global, and Ema was selected for Microsoft’s invite-only Pegasus Program. The company has raised $61M to date and says it will use the new funds to continue iterating its technology and expand investments in R&D and go-to-market. Ema offers an enterprise AI assistant sold as a platform with two products — Generative Workflow Engine (GWE) and EmaFusion — designed to emulate human responses and improve with usage and feedback. The company aggregates more than 30 large language models alongside its own smaller, domain-specific models in a patent-pending architecture to address accuracy, hallucination, and data protection concerns. Ema targets use cases across customer service (including technical support and tracking) and internal employee productivity, positioning itself as a cross‑use‑case solution rather than a single vertical tool. The startup emerged from stealth with paying customers it signed while private, including Envoy Global, TrueLayer and Moneyview. Leadership includes CEO and co-founder Surojit Chatterjee and co-founder/head of engineering Souvik Sen, both with extensive product and ML experience and numerous patents. The company says its goal is to free employees from routine work so they can focus on higher-value strategic tasks as it expands product applicability.