Ema Raises $77M to Let AI Agents Eat Enterprise Software Work

Ema has raised $77 million in a Series B round led by Bengaluru-based venture firm Creaegis, with existing investors Accel, Section 32, and Prosus increasing their stakes. The round was entirely primary equity — no debt, no secondary transactions — and brings the startup's total funding to $140 million. It also more than quadruples Ema's valuation from its last funding round in 2024, though the company declined to disclose the exact figure.

The funding lands as AI begins to compete for corporate dollars that businesses have long spent on enterprise software and IT services. Startups, major AI labs, and established software vendors are all fighting to capture that spending. Ema, founded in 2023 by former Google and Coinbase executive Surojit Chatterjee and ex-Okta executive Souvik Sen, both based in the Bay Area, is positioning itself squarely in that fight.

The startup deploys what it calls "AI employees" — systems that coordinate multiple AI agents to carry out multi-step business processes across a company's existing applications, rather than handling a single task at a time. Chatterjee argues that model will eventually cut companies' reliance on traditional software products, including software-as-a-service. Ema first wraps around an enterprise's existing applications, he said, before customers can reduce their dependence on some of those products — and in some cases replace them altogether.

"Many of our customers are already on the way to replace [large SaaS applications] completely, removing dependency on them, because they are mostly becoming like a database," Chatterjee told TechCrunch.

Not Afraid of the Frontier Labs

In recent months, major AI companies have pushed deeper into the enterprise market where Ema operates. Anthropic has expanded efforts to bring Claude into companies' core operations, including financial and legal work. OpenAI has established teams of forward-deployed engineers who work alongside customers to put AI into production.

Chatterjee does not see the frontier labs as direct competitors. Ema's software can draw on more than 150 models, including frontier and open source models, while the startup focuses on the domain knowledge, integrations, and orchestration needed to automate business processes end-to-end.

"Progress in frontier models is actually very beneficial to us," Chatterjee said.

The numbers suggest traction. Ema has more than 50 active enterprise deals and over 1 million active enterprise users, and it has handled more than 5 million actions and queries. Its customers include NTT DATA, Hitachi, ADP, PwC, Google, KPMG, Wipro, and Microsoft.

Revenue has grown 50-fold over the past two years, and revenue bookings have surpassed $150 million. Chatterjee cautioned that the bookings figure includes the total value of multiyear contracts — two- and three-year deals — rather than representing annual recurring revenue. He declined to disclose the company's current annualized revenue run rate.

Expansion inside existing accounts is running hot. More than 90% of Ema's customers have expanded beyond their initial use case, with some deploying the technology across dozens of workflows, according to Chatterjee. The startup's net dollar retention rate sits around 180%, meaning existing customers are spending substantially more with Ema over time.

Services Firms Are Rethinking Their Own Models

Ema is also looking beyond software itself. Chatterjee said AI can take over some of the implementation, integration, and consulting work that companies have traditionally paid IT services firms to perform around enterprise software. Rather than resisting, some of those firms are partnering.

"A lot of the services companies are working with us," Chatterjee said. "They are also dramatically changing or disrupting their own business models because they understand the human-forward model may not be the best model going forward."

Despite taking on work traditionally handled by software and services providers, Ema has maintained gross margins close to 80%, Chatterjee said. The startup requires less human support as its AI systems learn from deployments, which should help improve margins over time.

Its pricing model breaks from both seat-based software licensing and token-based AI metering. Ema charges customers based on the completion of tasks and business outcomes.

Much of the new capital will fund go-to-market expansion, particularly sales and marketing, after the company spent its first years largely building the product. The Mountain View-headquartered startup has grown to nearly 200 employees and maintains offices in Bengaluru, London, and Vancouver.

Ema has focused primarily on U.S. and European customers to date. Over the next year, it plans to push into new markets across Asia-Pacific, South America, and parts of the Middle East — betting that outcome-based pricing and agent orchestration, not seat licenses, will define the next era of enterprise software spend.