Orbi, a startup that builds artificial intelligence (AI) agents to automate repetitive and tedious tasks for businesses, has raised $30 million (about 42 billion won) in funding. The corporate’s AI agent is characterised by its use of enormous motion models (LAMs) that may “watch and learn” to handle more complex and diverse tasks than conventional RPA (robotic process automation).
VentureBeat reported on the twenty seventh (local time) that startup Orbi attracted $30 million in Series A investment for its enterprise AI agent constructing platform.
This investment was participated by Latest Enterprise Associates, Wing Enterprise Capital, WndrCo, and Fair Enterprise Capital.
Orbi AI goals to unravel complex enterprise task automation problems using advanced generative AI models. It has unveiled a brand new ‘Large Motion Model (LAM)’ technology that allows AI agents to learn and automate complex business processes.
This can be a GPT type model, which implies the model takes all observed actions as input and outputs the code required to automate tasks.
Existing automation tools are good at performing single tasks equivalent to data entry, document processing, and verifying complex forms and contracts, but they struggle to handle complex workflows. It is because existing tools are designed to perform “rules-based automation,” which applies human-created rules to store, organize, and manipulate data.
One in all the issues with such rule-based automation is its lack of flexibility, which implies it might probably’t handle tasks when technology or formats change. For instance, if a rule-based automation tool is tasked with transferring data from one interface to a different, it’s likely that it won’t have the opportunity to finish the duty if there are differences in the info format.
Due to this fact, Orbi developed an end-to-end generative AI platform that mixes large motion models and AI agents to transcend rule-based automation.
The platform trains LAM using multimodal inputs equivalent to screenshots, clicks, and keystrokes. The AI agent then generates scripts and code to automate workflows, while handling exceptions with human input for feedback.
An auditor at a Fortune 500 company shared how he uses the Orbi platform to audit thousands and thousands of employee-submitted receipts, a task that typically takes 10-Quarter-hour per receipt.
But once you put in Orbi, you simply have to point out what you are doing. With only one iteration, Orbi can generate a workflow with code. For very complex cases, you’ll be able to ask a human auditor for feedback.
This platform known as ‘APA (Agentic Process Automation)’ and is described as one step more advanced than traditional RPA tools.
When an AI agent receives a task to perform, including instructions, it autonomously creates a workflow. This workflow integrates with other specialized AI agents to perform various subtasks required to perform the duty. These subtasks may include data evaluation or customer interactions. The reason is that through the use of various AI agents to finish related tasks, complex tasks could be handled in various scenarios.
“Existing RPA systems weren’t designed to unravel dynamic and sophisticated enterprise workflows,” said Bella Liu, CEO of Orbi. “In contrast, we’re proving what AI can do for enterprises.”
“The platform has the power to look at and learn from every motion you are taking,” he continued. “It analyzes the context and creates automated workflows that could be executed immediately.”
Orbi said it would use the brand new funds to bring its AI process automation platform to market and advance its development.
Reporter Park Chan cpark@aitimes.com
