Ema raises $77M on accounts that spend 80% more each year

Five million actions and queries divided by more than a million active enterprise users is about five actions each, across the life of the deployments ^[1]. Averaged over 50-plus enterprise deals, the same user count gives roughly 20,000 active users per account ^[6]. Those are reach numbers. Depth shows up in two other figures Chatterjee gave TechCrunch: more than 90% of customers have gone past their first use case, some into dozens of workflows, and net dollar retention is around 180% ^[16][17].

Net dollar retention of 180% means a group of customers paying $100 a year ago is paying about $180 now, after churn and downgrades, an 80% net expansion ^[3]. For a vendor that does not sell seats ^[21], that figure says more than a user count does, because an enterprise user who logs in once and never comes back costs the customer nothing.

Chatterjee said the $150M bookings figure is the total value of multiyear contracts, including two- and three-year deals, and not annual recurring revenue ^[14][15]. Spread evenly, $150M is $50M a year over three years, or $75M a year over two ^[4]. He declined to give the current annualized run rate ^[15]. Ema had raised $63M in total before this round ^[2].

The case for eating into software budgets is Chatterjee's own. He said many of Ema's customers are already on the way to replacing large SaaS applications completely and removing their dependency on those products, because they are mostly becoming like a database ^[7]. The order he described starts with addition: Ema wraps around an enterprise's existing applications, and customers reduce their dependence on those products afterwards, in some cases replacing them ^[8].

The services half is spending that is already going out the door. NTT DATA, Hitachi, ADP, PwC, KPMG and Wipro are named customers ^[13], and implementation, integration and consulting around enterprise software is the work Chatterjee said AI can take over from services firms ^[18]. "A lot of the services companies are working with us," he 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" ^[19]. Gross margins are close to 80%, he said, with less human support needed as the systems learn from deployments ^[20]. That leaves about $20 of delivery cost per $100 of revenue ^[5].

The reporting does not say which budget line the deals are funded from. For whoever signs year two, the test has two parts: name the license or the statement of work that gets smaller, and name the month it gets smaller. Ema's own sequence makes the year-one answer "nothing yet." If year two gets the same answer, the deployment is sitting on top of the software and services budget, and it renews at about 1.8 times its first-year size ^[3].