Ema AI employees just picked up a $77 million vote of confidence from investors who believe large companies are done testing AI agents and ready to put them to work. The Mountain View startup closed a Series B round led by Bengaluru-based Creaegis. Existing backers Accel, Section 32, and Prosus all returned with significantly bigger checks than they wrote the first time. The round brings Ema’s total funding to $140 million and comes just two years after the company launched out of stealth. The company says its valuation has more than quadrupled since its last raise in 2024, though it has not disclosed the exact figure. Every dollar arrived as primary equity. No debt was involved, and no existing shares changed hands.
How Ema AI Employees Actually Get Work Done
Most AI products sold to enterprises still work like a smarter search bar. Ask a question, get an answer, move on. Ema built something different. Its agents are grouped into what the company calls AI employees, systems designed to complete entire workflows inside a company’s HR, IT, and finance departments rather than respond to a single prompt at a time. Co-founder and CEO Surojit Chatterjee, who previously held senior roles at Google and Coinbase, has been direct about where this leads. He said many of Ema’s customers are already working to remove their dependence on large software platforms entirely, describing that software as becoming little more than a database sitting behind the agents actually doing the work.
That is a significant claim to make at a time when the companies building the underlying AI models are entering the same market Ema operates in. Anthropic has pushed its Claude models deeper into corporate finance and legal functions this year. OpenAI has built teams of engineers who work directly inside client companies to get its models running in production. Chatterjee does not see either as a threat. Ema’s platform pulls from more than 150 different AI models, he said, and progress at the frontier labs strengthens his product rather than competing with it, since the value his company sells sits in workflow design and integration rather than in the model doing the reasoning.
The growth numbers explain why investors were willing to pay four times the previous valuation. Ema now serves more than 50 active enterprise customers, including Google, Microsoft, PwC, KPMG, ADP, NTT DATA, Hitachi, and Wipro. Its agents have handled over 5 million actions and queries across more than a million active users. Revenue has grown fifty-fold in two years, and contracted bookings from multi-year deals have crossed $150 million. What stands out more than the top-line figures is customer behavior after signing. More than 90 percent expand beyond their original use case, and net dollar retention sits near 180 percent, meaning existing clients are nearly doubling their spending over time.
Two examples make the idea concrete. At Wipro, an Ema-powered assistant now supports more than 240,000 employees across 65 countries. It handles roughly 2.9 million queries a year through close to 100 automated workflows and has cut IT support ticket volume in half. At Hitachi, the deployment moved from concept to full production in under four weeks, connecting to more than 20 internal systems. Support tickets there have dropped 30 percent, and the company reports a 70 percent gain in staff efficiency. Neither company replaced its entire software stack. Both show how quickly an AI employee can absorb work that used to sit with a help desk or an outside consultant.
That shift extends beyond software budgets into the consulting world. Chatterjee said several IT services firms are now partnering with Ema rather than competing against it, having concluded that billing hourly for human consultants may not hold up much longer. It fits a pattern showing up across enterprise technology this year, where large contracts are being rebuilt around automated systems instead of manual processes. HCN recently covered a similar shift in Airbus’s cybersecurity partnership with France, where a traditional vendor relationship was restructured around automated monitoring rather than a fully human-run operation.
Ema has managed to keep gross margins near 80 percent despite taking on work that would normally require large teams of contractors. Company executives credit that partly to how little human oversight its systems need once a deployment matures. The pricing model reflects the same logic. Ema does not charge by software seat or by AI token usage. It charges based on whether a task actually gets completed, according to TechCrunch’s reporting on the funding round.
Most of the new capital will go toward something Ema has largely avoided until now: building out sales and marketing. The company has spent its first two years focused almost entirely on the product itself and has grown to roughly 200 employees across offices in Mountain View, Bengaluru, London, and Vancouver. Its customer base has so far been concentrated in the United States and Europe, but the company plans to expand into Asia-Pacific, South America, and parts of the Middle East over the coming year.
Whether Ema’s AI employees end up replacing enterprise software altogether, or simply become another costly layer running on top of it, remains an open question. It is one that Ema, its competitors, and the software vendors watching from the sidelines are all still working out in real time.
[…] agent builders. On Sept. 23, Ema, which builds AI employees for HR, IT and finance work, raised $77 million in a Series B round, as HCN has reported. The company says its platform links to more than 250 business systems and […]