You.Com AI Agents Aim to Increase Enterprise Productivity

Since launching in November 2021 to challenge Google’s search monopoly, you.com has served 1 billion queries, boasts millions of active users, and grown its ARR by 500% since January 2024. Today, it announced a $50 million Series B round led by Georgian, with participation from SBVA (formerly Softbank Ventures Asia), Salesforce Ventures, NVIDIA, DuckDuckGo, and Day One Ventures, bringing its total funding to $99 million.

Today, you.com also marks a new phase in its evolution, moving beyond the individual worker mode toward team-centric collaboration. “We’re almost exclusively focused on the enterprise, calling ourselves the productivity engine now,” says you.com co-founder and CEO Richard Socher. Enterprise teams will be able to build custom AI agents on any AI model for any task, exchange chat threads, and access unlimited file transfers without worrying about data storage. In addition, enterprises can integrate you.com APIs to enhance their products with AI and information from the web.

A recent McKinsey survey of executives implementing generative AI found that their biggest concern was accuracy. In fact, it was the only risk that respondents were significantly more likely than last year to say their organizations were actively working to mitigate. Nearly a quarter of respondents have experienced negative consequences from generative AI inaccuracy.

Accuracy, Socher says, is what sets you.com apart in a crowded market of AI agents. Like self-driving cars, Socher says, they typically fail at the “last mile,” failing to “reliably, at scale” do what they’re asked to do and what they’re told to do. AI agents in the enterprise need to be able to reliably and accurately answer complex queries, “10 times better, effectively replacing up to 10 different queries.”

Over the past few years, the you.com team has invested much of its collective intelligence and expertise in search/NLP/LLM into honing its ability to provide knowledge workers with exactly the kind of help they need to perform specific tasks and collaborate effectively with others. “Knowing how to use Google search,” says Socher, “was an advantage a quarter of a century ago. Knowing how to prompt engineers and build your own custom agents for all kinds of workflows is an advantage today.”

Achieving the Holy Grail of AI gen, i.e. hallucination-free experience, and earning the trust of knowledge workers is the ambition of you.com. To achieve this, it combines — and continually improves — a number of technological foundations. These include having a solid search backend; being model-agnostic and knowing which LLM model is right for the job, given its relative advantages (some models are better at reasoning, others at programming, etc.); access to real-time networks and citations from up-to-date, verifiable, and authoritative sources; and understanding intent, enabling query adaptation and dynamic adjustment of prompts, models, and techniques to fit the task at hand.

Combining these tools on a single platform creates AI-powered enterprise productivity agents that perform tasks such as searching, researching, coding, creating with or without manual prompts, and these tasks can be shared with colleagues.

According to CB Insights, “the landscape of VC-backed AI agent startups is dominated by a focus on horizontal applications—sales, customer service, and other workflows related to enterprise and general productivity.” As of July 29, 2024, these startups have raised $1.55 billion since 2022 (excluding OpenAI’s funding). There’s been a “surge in investment and deal levels” recently, with $980 million raised across 40 deals this year. Google announced a “Universal AI Agent,” and Microsoft and Amazon are developing AI agents focused on automating software workflows.

Early examples of the impact of AI agents on enterprises include Klarna’s AI assistant, which completed 2.3 million conversations in one month of operation, or two-thirds of Klarna’s customer service chats, performing the work equivalent of 700 full-time agents; Amazon Q, Amazon’s next-generation AI assistant for software development, which updates an application to Java 17 in just a few hours, a process that would normally take 50 development days, saving the equivalent of 4,500 developer years of work; and Walmart using multiple large language models to accurately create or enhance over 850 million data items in its catalog, which would require almost 100 times more work than it currently does to complete in the same time.

You.com’s AI agents can help “knowledge workers be more productive, whether it’s through fast, accurate answers, research and analysis, problem-solving, or content creation,” Socher says.