AI “agents” are generative AI models that may perform actions on their very own, equivalent to copying information from an email and pasting it into a spreadsheet. compressors that increase efficiency. This could also be a bit premature given the tendency of models to make mistakes. But at least some of the founders (i.e analysts AND investors) seem convinced that agents represent the next breakthrough in generative AI.
Bella Liu and William Lu are two such founders. Their company, AI Orbsis building a generative AI platform that goals to automate a range of various business processes, including processes involving data entry, document processing and form validation.
Many startups offer tools to automate repetitive, monotonous business processes on the back end (see Parabola, Tines, Induced AI powered by Sam Altman, and Tektonic AI, to call a few). Incumbents equivalent to Automation Anywhere and UiPath have also moved to AI as they fight to maintain pace with the competition in generative AI.
But Liu and Lu say Orba’s technology stands out for its ability to learn and reply to workflows in real time and understand patterns and relationships in an enterprise’s unstructured data.
“The Orby platform watches how employees perform their work to automatically create automations for complex tasks that require a certain level of reasoning and understanding,” explained Liu, CEO of Orby. “An AI agent installed on an employee’s computer effectively observes, learns and generates automations, adapting the model as it learns.”
Liu and Lu say that for Orba, which is set to launch in 2023, they desired to create AI that might understand some of the decisions that low-level employees make and abstract them from those decisions, freeing up employees to focus on more essential matters.
Liu previously led AI and automation efforts at IBM, including AI product planning and M&A, and was UiPath’s director of AI product management. Lu is a former Nvidia systems engineer who joined Google Cloud as an engineering manager, helping design generative AI and database extraction technology.
Orby’s supposed secret sauce is a cloud-based generative AI model that’s tuned to handle customer tasks like checking expense reports. The model relies in part on symbolic AI, a type of AI that uses rules, like mathematical theorems, to reason about solutions to problems.
Symbolic AI itself could be stiff and slow, especially when dealing with large and complex data sets. To work well, it requires clearly defined knowledge and context. However, recent research has shown that it is scalable when combined with traditional AI model architectures.
“We have been designing this AI model for the past two years and have conducted successful tests,” Liu said. “There are few pure generative AI corporations that attack enterprises from end to finish with something end-to-end. We are one of them.”
Liu says Orby’s model can intelligently adapt to changes in workflows, equivalent to when an app’s user interface is updated, by analyzing API interactions and worker browser usage. Having software that monitors an worker’s every move seems like a privacy disaster waiting to occur. But Liu says Orby doesn’t store most of the customer data; it only uses certain telemetry data to enhance its model, encrypting data each in transit and at rest.
“People are completely trapped in a feedback loop,” she added.
Orby, which recently raised $30 million in a Series A financing round co-led by New Enterprise Associates, WndrCo and Wing (sources put a post-money valuation of $120 million), competes in a difficult sector. Upcoming agentic AI from generative AI powerhouses equivalent to OpenAI and Anthropic has dampened the prospects of each incumbent and smaller players.
Adept, a startup building AI agent technology focused on enterprise applications supposedly on the verge of a Microsoft acquisition deal before it could even deliver a single product. Amazon and Google have released AI agent tools to little traction. Elsewhere, UiPath, despite a surge in generative AI initiatives last 12 months, saw sales plum in the last fiscal quarter.
Liu says Orby can get ahead by taking a systematic approach to market entry. He says the company already generates revenue from about a dozen customers and plans to make use of $35 million of its war chest to expand its team of about 30 based in Mountain View.
“These funds are being used to scale our go-to-market, customer service, product and technical organizations,” she said. “The enterprise market has an insatiable appetite for generative AI solutions that demonstrably improve business outcomes; they are just trying to figure out where best to apply this technology in the near future before they apply it across their entire business.”