Generative AI has a firm grip on public technology discourse today. A brand new startup called Ema of San Francisco believes that is rather more than simply a passing fancy. Today, the corporate is emerging from obscurity with an eponymous product that expects to open a brand new chapter in how AI, particularly generative AI, will change the way in which we work.

“Our goal is to construct a universal AI workforce,” said Surojit Chatterjee, CEO and co-founder, in an interview. “Our goal is to automate the mundane tasks that employees in every company perform each day… giving them more time for more precious and strategic tasks.”

The company and its investors are putting money and revenue behind their words: It has already raised $25 million from a formidable list of backers, together with customers it quietly recruited while still within the shadows , to refute all allegations of vaporware, including Envoy Global, TrueLayer and Moneview.

What Ema can do is allow these corporations to make use of it in applications starting from customer support – including providing technical support to users in addition to tracking and other functions – to internal productivity applications for workers. Ema’s two products – Generative Workflow Engine (GWE) and EmaFusion – are designed to “emulate human reactions” but evolve with feedback as use increases.

As Chatterjee describes it, it isn’t just robotic process automation (that is so 2010s) and it isn’t just AI to hurry up certain tasks (that goes back even further), and it isn’t just one other generation of AI accuracy waiting to fail mocked on social media.

Chatterjee says Ema – which is an acronym for Enterprise Machine Assistant – draws on greater than 30 large language models and combines them with its own “smaller, domain-specific models” in a patent-pending platform “to deal with these.” all the issues you’ve seen with accuracy, hallucination, privacy, etc.”

This early round adds lots of names to Ema’s cap table. Accel, Section 32 and Prosus Ventures are co-leaders, with Wipro Ventures, Venture Highway, AME Cloud Ventures, Frontier Ventures, Maum Group and Firebolt Ventures also participating. There are also some notable individual supporters: including Sheryl Sandberg, Dustin Moskovitz, Jerry Yang, Divesh Makan and David Baszucki.

Currently, there are already dozens, perhaps lots of, of corporations developing GenAI tools for enterprises, each those working on solutions for specific industries or use cases and impressive, self-led style shifts like Ema’s. If you are wondering why this particular GenAI startup is attracting these investors’ attention, it might be partly because they’re already doing business. But it also relies on the background of the team.

Prior to Ema, Chatterjee was Chief Product Officer of Coinbase in its run-up to its IPO. Previously, he was VP of Product at Google for each the mobile ads and shopping businesses. He himself has around 40 patents in areas akin to enterprise software for machine learning and promoting technology.

The other co-founder, Souvik Sen, Ema’s technical director, has equally impressive experience. Most recently, he was VP of Engineering at Okta, where he led data, machine learning and devices. He also previously worked at Google, where he was technical lead for data and machine learning, specializing in privacy and security. He also owns 37 patents.

The combined experience of those two adds weight to the corporate’s ambitions and the likelihood of achieving them. But many details that would definitely play a job in development are also omitted.

For example, consider Chatterjee’s expertise in e-commerce and adtech. Given that these are cornerstones of the way in which so many corporations interact with customers today, it seems inevitable that they are going to play a job in how Ema might evolve if successful.

On the opposite hand, if there may be a founder who has previously needed to worry about data protection and privacy, the startup can have a greater likelihood of not screwing up these issues. At least we are able to hope! It is AI, in any case, and it’s a Silicon Valley startup that may ultimately concentrate on the business at hand and the usage of technology to realize it.

What’s notable in the mean time is that ambitious startups are working to construct products that span across different LLM silos to realize more advanced outcomes. This is maybe an early sign that the LLMs are more interchangeable than you’d assume over time, and in addition more standardized.

And the power to cover different use cases gives the startup potential diversification that would help grow its overall business and utility, investors say.

“Most point-generation AI solutions provide high value for specific use cases, but are difficult to increase either across use cases and even adjoining use cases, and more importantly, large enterprises are concerned about fragmentation and access to their sensitive data through so many alternative applications,” Ashutosh Sharma, head of investment at Prosus Ventures in India, told TechCrunch. “Ema can solve these problems and deliver high accuracy with optimal return on investment.”

This article was originally published at