Orby Develops AI Agents Tailored for Enterprise Use

AI "agents" are generative AI models designed to perform tasks autonomously, such as transferring information from an email into a spreadsheet. While these agents are praised as productivity enhancers, some believe this acclaim may be premature due to their tendency to make errors. Nonetheless, certain founders, analysts, and investors see AI agents as the next major advancement in generative AI.

Bella Liu and William Lu are among these founders. They have established Orby AI, a generative AI platform aimed at automating diverse business workflows, including data entry, document processing, and forms validation.

Numerous startups provide tools to automate repetitive back-office operations, such as Parabola, Tines, and Sam Altman-backed Induced AI, along with established companies like Automation Anywhere and UiPath, which are embracing AI to keep pace with emerging generative AI solutions.

However, Liu and Lu assert that Orby’s technology differentiates itself through its capability to learn and adapt to workflows in real time while recognizing patterns within an enterprise’s unstructured data. "Orby’s platform observes how employees work to automatically develop automations for complex tasks requiring reasoning and understanding," explained Liu, the CEO of Orby. "An AI agent on a worker’s computer learns and creates automations, continually refining its model as it gathers more information."

Launched in 2023, Orby aims to create an AI that comprehends the low-level decisions workers make, allowing them to concentrate on more complex tasks. Liu previously led AI and automation initiatives at IBM and UiPath. In contrast, Lu is a former Nvidia systems engineer who worked at Google Cloud, where he focused on designing generative AI technologies for document and database extraction.

Orby’s unique advantage lies in its cloud-based generative AI model, tailored to fulfill specific customer tasks such as validating expense reports. This model utilizes symbolic AI, a type of AI that employs rules, like mathematical theorems, to derive solutions.

While symbolic AI can struggle with large, complex datasets due to its need for well-defined knowledge and context, recent studies indicate it can scale effectively when integrated with traditional AI model architectures. "We’ve been refining this AI model for two years and have conducted successful trials," Liu noted. "Few generative AI companies are addressing enterprise needs with an end-to-end solution. We are among them."

Liu emphasizes that Orby’s model can adapt intelligently to evolving workflows, including changes in application user interfaces, by analyzing API interactions and user behavior. Although monitoring an employee’s actions raises potential privacy concerns, Liu assures that Orby does not retain most customer data. Instead, it uses select telemetry data to enhance its model, ensuring data is encrypted both in transit and at rest. “Humans are kept completely in the feedback loop,” she added.

Recently, Orby raised $30 million in a Series A funding round co-led by New Enterprise Associates, WndrCo, and Wing, bringing its post-money valuation to $120 million. The company is navigating a competitive landscape, as major generative AI players like OpenAI and Anthropic are developing their own AI agents, affecting smaller competitors.

Additionally, Adept, a startup focused on enterprise AI agents, is reportedly nearing an acqui-hire deal with Microsoft before launching a product. Amazon and Google have introduced AI agent tools with minimal impact, and UiPath is experiencing a decline in sales despite ramping up its generative AI endeavors over the past year.

Liu believes Orby can succeed through a systematic go-to-market strategy. The company already generates revenue from about a dozen customers and intends to invest its $35 million in growth by expanding its 30-person team based in Mountain View. "The funds will help us scale our go-to-market strategy, customer support, product, and technical teams," she said. "Enterprise clients are eager for generative AI solutions that clearly enhance business performance, and they're currently exploring the best ways to implement this technology before rolling it out company-wide."