An artificial intelligence startup founded on Long Island has landed millions of dollars in its latest venture capital round and has launched a limited rollout of a tool that can turn a web browser into a digital butler, the company’s 23-year-old chief executive said.

OthersideAI has attracted $5.4 million in two venture capital rounds since it was started in 2020, with additional funding yet to be announced, said Matt Shumer, who is also a co-founder.

The latest funding round, disclosed in March, totaled $2.8 million from previous investors Madrona Venture Group LLC, Active Venture Partners LLC and others. That gives the company an implied valuation — based on what investors were willing to pay for shares in the latest funding round — of about $50 million.

“We’re well capitalized,” Shumer said of his seven-person company. “We have well over one million users. Revenue is in the millions.”

What the AI agent will be able to do:

  • Book a flight to Paris
  • Order a pepperoni pizza
  • Organize a Gmail inbox

Source: OthersideAI

Shumer was a business and computer science major who left Syracuse University in his sophomore year of 2020 to start OthersideAI in his parents’ Melville home.

The company’s first commercial product, HyperWrite, works as an extension on the Chrome browser to perform tasks like expanding a few bullet points into a full-blown email, responding to emails in your writing style, creating newsletters, writing the introductory paragraph of an academic essay and summarizing a text or editing it for clarity, tone and style.

HyperWrite is free for limited use and offers monthly subscriptions for $19.99 and $49.99.

The rollout of HyperWrite predates OpenAI’s release in November 2022 of ChatGPT, a conversational chatbot that has captured the public imagination about the capabilities of AI.

ChatGPT was schooled on large-language models — including books and information from the Internet — that give it a breadth of knowledge in topics as diverse as law and medicine.    

Widespread interest in ChatGPT also has fueled an AI gold rush.

OpenAI faces rivalry from tech giants like Google’s parent company Alphabet as well as early-stage companies like Toronto-based Cohere.   

Aiden Gomez, Cohere co-founder and CEO, is a veteran of Alphabet’s AI research program. While at the Google Brain unit, he co-authored a seminal paper called “Attention is all you need,” laying out the architecture of new-generation AI systems.

Gomez and other Cohere executives are investors in OthersideAI.


Among the requests OthersideAI’s  personal assistant can tackle, according to the company web site:

  • Organize my Gmail inbox and draft responses;
  • Book me a flight from New York to Paris;
  • Order a large pepperoni pizza;
  • Find engineering candidates on LinkedIn.

Shumer, however, acknowledged that the personal assistant — which he said is the first of its kind — also makes mistakes.

Concerns about safety and accuracy are prompting OthersideAI to limit the rollout, monitor how the system is operating and build guardrails for the early testers, who have a tool allowing them to verify credit card transactions before they are submitted.

“We could roll this out faster… but my view is that it’s not the right thing for society,” Shumer said.

Niranjan Balasubramanian, a faculty member at Stony Brook University’s Institute for AI-Driven Discovery and Innovation, said automating complex tasks remains a hurdle for these systems based on large-language models.

“Are they perfect? Not by a stretch,” he said. “The models are known to make mistakes in the individual steps. Sometimesthey make up stuff.”

For instance, Balasubramanian said, a traveler might ask a digital agent to find the cheapest flight to a destination, but the system might book a flight with a three-day layover.

“If you let them loose with your credit card… things can quickly go haywire,” he said.

Jason Kuperberg, OthersideAI co-founder, said in a LinkedIn post that by rolling out the personal assistant to select users, the company will be able to make safety improvements before a wider release.

“We recognize the immense power and responsibility that comes with developing AI agents and are committed to handling it responsibly,” he said.

Source link