There is a surge of startups delving into the world of enterprise AI assistants. However, what sets Narada AI apart is its ability to perform tasks across multiple work apps simultaneously. This startup, inspired by fresh research from UC Berkeley, has been in stealth mode for two years and just made its public debut at Startup Battlefield 20 during TechCrunch Disrupt 2024.

Co-founded by Kurt Keutzer and Amir Gholami, Narada is based on their research on “LLM Compilers,” AI systems capable of executing multiple functions at once. Park, the startup’s CEO and co-founder, emphasized their unique approach as a key differentiator in the realm of AI chatbots.

The AI assistant by Narada operates efficiently by carrying out various tasks using generative AI across different work apps. From drafting emails to creating calendar invites, taking meeting notes, and browsing the web, this assistant is designed to streamline your daily routines.

To navigate without APIs, like in SAP or Salesforce, Narada’s assistant uses a front-end approach called Web Redemption. This method, according to Gholami, Narada’s CTO, is akin to a Roomba mapping out new websites or applications to comprehend and utilize them effectively.

While Narada is not alone in this AI assistant competition, its focus on enterprise applications sets it apart. The startup has already caught the attention of a Fortune 500 company, showcasing the potential of its agent in the corporate world.

Although Narada isn’t a replacement for a real-life assistant, it provides a convenient shortcut for routine tasks. However, granting such access to an AI assistant may raise concerns about data privacy and security. Nonetheless, Narada assures that it does not use customer data to train its AI models.

Having raised a few million dollars from advisers, Narada is now seeking additional funding from traditional VCs to further its innovative AI assistant technology.