The AI start-up, which boasts of an impressive line-up of investors such as Accel, Sec 32 and Prosus Ventures, has also caught the fancy of industry heavyweights.
Trustworthy and capable, she can juggle roles such as data scientist and customer service and move on to specialised tasks like pharmacist’s assistant. ‘Universal AI employee' Ema can morph into any role and has what it takes to quickly become an organisation’s best worker, says founder Surojit Chatterjee.
Chatterjee, who previously spearheaded Google Ads evolution, teamed up with Souvik Sen, builder of Okta’s and Google’s largest ML systems, for the San Francisco-based AI start-up. The company is Ema and so is the product designed to match the capabilities of a human.
"Our goal at Ema is to help automate most of the repetitive and tedious tasks that human employees perform today and free them up to do more valuable work in the enterprise," Chatterjee told PTI.
Ema, which can act as a conversational agent, as a chatbot and as a workflow automation tool all in one, has already been hired for multiple roles by many enterprises, he added. These include Envoy Global, TrueLayer, Moneyview, and Exclusive Networks.
The buzzing AI startup, which boasts of an impressive line-up of investors such as Accel, Sec 32 and Prosus Ventures, has also caught the fancy of industry heavyweights and high profile names with Sheryl Sandberg, Dustin Moskovitz, Jerry Yang and others backing it.
The founders envision Ema to be a "force for good" - aiding where companies can't hire customer support agents fast enough or helping out firms grappling with seasonally large volumes of support tickets so employees can take on more strategic and creative functions.
"We believe in a future where human and AI employees will work hand-in-hand and usher in a new era of prosperity," Chatterjee said.
Excerpts from the interview:
Q: What sets Ema's Universal AI Employee apart from other solutions currently available in the market?
A: Our goal at Ema is to help automate most of the repetitive and tedious tasks that human employees perform today and free them up to do more valuable work in the enterprise. Ema has achieved a few significant technological breakthroughs.
Ema opens up a new category in Enterprise Gen AI space called Universal AI Employee. Ema can act as a conversational agent, as a chatbot and as a workflow automation tool all in one. Ema can morph into taking on any role in the organisation - from standard roles like customer service, data scientist, sales assistant to more specialised roles such as a ‘pharmacist’s assistant’ to help make faster decisions on pre-authorisations.
Ema is simple to onboard and use. She is highly trustworthy, extremely accurate and is capable of quickly becoming one of your best employees.
Based on evaluation by human experts appointed by our customers, Ema is already performing at or above human performance resulting in dramatic gains in productivity for them.
Three aspects that differentiate Ema from other solutions currently available in the market -- it is simple, trusted, and accurate.
With a simple conversation, you can activate a new AI Employee, a persona if you may, capable of performing standard or specialised tasks for your enterprise. These personas collaborate seamlessly with human colleagues and can automatically generate the workflows needed to complete tasks in a given role.
We mitigate privacy and security risks by using robust data protection and security algorithms. We are fully compliant with the top international standards - SOC2, HIPAA, GDPR, and ISO27001.
Further, we have created EmaFusion, our unique model with over 2 trillion parameters, by combining over 30 LLMs and various smaller ML models we've developed internally. EmaFusion maximises accuracy at the lowest possible cost for most common tasks in the enterprise.
Q: Given the increasing scrutiny over AI ethics and bias, can you elaborate on the measures Ema is taking to ensure its AI Employees operate ethically and how you are addressing potential biases in AI outputs?
A: Ema has a unique three-pronged approach to mitigate bias and operate ethically. It is powered by our proprietary 2 trillion plus parameter model, EmaFusion, that intelligently blends 30 plus LLMs (including GPT-4, GPT-3.5, Mistral, Claude, Llama2 etc.) and many smaller models we built in-house. Cross-checking answers with multiple expert models helps us dramatically reduce bias.
Secondly, when we are choosing models to put into EmaFusion, we are highly vigilant and choose only models with strict ethical standards.
Additionally, we put safeguards to make sure Ema is behaving ethically in the enterprise. For example, Ema can automatically check for any copyright violations when she is generating content.
Finally, Ema is leading the industry in building explainability of Gen AI for the enterprise. Ema shows her work to human supervisors who can quickly check and give her feedback. Ema learns from such conversational feedback and can adapt her behaviour quickly. Before any deployment, Ema goes through rigorous training, testing and human evaluation from experts inside the enterprise. This further reduces the possibility of bias in Ema’s actions.
Q: There is growing fear about AI and automation taking away jobs. How does Ema envision its Universal AI Employee fitting into the broader context of work?
A: Like every major technological change in the past - from industrial revolution to the internet - we believe Gen AI will create more jobs and more wealth for everyone and will be a key driver for productivity growth in the future. We envision Ema to be a force for good by reducing tedious tasks in the enterprise so every employee can focus on more strategic, creative and purposeful work.
These workforce transitions are inevitable - enabling an entirely new class of work and new businesses that weren't possible 10 years ago. This is what fuels business and economic growth.
For example, many of our customers were struggling because they could not hire customer support agents fast enough to support their growth or were impacted by seasonally large volumes of support tickets that negatively impacted their service quality. With Ema helping their customer support team, they can keep pace with their customer growth and improve customer satisfaction.
In fact, with higher revenue and profits, they are now able to hire more people in other business areas. This is particularly important for small and medium size businesses which are resource constrained.
Ema’s significant impact on productivity helps create financial resources to re-invest in business growth. We believe in a future where human and AI employees will work hand-in-hand and usher in a new era of prosperity.