Ema raises $77M for AI employees and prices by outcomes, not seats

The most interesting number in Ema's funding news is not the $77 million. It is how they charge.

What happened

Ema, a startup that uses teams of AI agents to automate processes across HR, IT and finance, raised a $77 million Series B led by Creaegis, bringing total funding to $140 million, TechCrunch reports. Its "AI employees" coordinate multiple agents to run multi-step processes across a company's existing applications.

CEO Surojit Chatterjee told TechCrunch that Ema does not charge by seats or tokens. Pricing is tied to completed tasks and business outcomes. The platform can draw on more than 150 models, and more than 90% of customers have expanded beyond their first use case.

SiliconANGLE adds that the agents plan a task, execute it inside existing systems, check their own work and notify humans when approval is needed for sensitive actions. Wipro has deployed an Ema-powered assistant for more than 240,000 employees, handling around 2.9 million queries a year across about 100 workflows, with a 50% reduction in IT support tickets. Hitachi reportedly went from concept to production in under four weeks across more than 20 enterprise systems.

Ema plans to expand into Asia-Pacific, South America and parts of the Middle East over the next year.

My take

This is enterprise scale, but the lessons carry down to a 20 person business:

1. Wrap, do not replace. Ema starts by wrapping around the apps a company already uses. For smaller teams that means building agents on top of the CRM and helpdesk you have, not migrating.
2. Approval gates are a feature. Agents that pause for human sign-off on sensitive actions are how you get a finance or HR team to say yes.
3. Measure outcomes. If an enterprise vendor can price per completed task, you can at least measure your automations that way: tickets closed, invoices matched, leads routed. That is the number your boss or client cares about.
4. Start narrow, then expand. The 90% expansion stat says the same thing I see in client work. One workflow that works earns the next five.

I design and build AI automations that do real jobs like lead routing, reporting, data entry, IT and HR request handling, and multi-step agent workflows. See the range at romielwillautomate.dev.