Automating work with Orby AI

The average enterprise now runs hundreds of software-as-a-service (SaaS) applications. Okta's 2023 Businesses at Work report puts the figure at 130 per large organization, up roughly 46% from 89 in 2018, and the number keeps growing. Each new tool adds capability, but it also adds friction. Employees toggle between apps, copy data from one system to another, and manually orchestrate workflows that span 5, 10, or 15 different platforms. The result is a productivity cost: humans spend hours each week serving as manual integrators between disconnected systems.

We believe Large Action Models (LAMs) represent a step-change from legacy robotic process automation (RPA). Where traditional RPA records brittle, user-interface-based scripts that break when a button moves, LAMs observe how people actually work, learn the underlying intent, and automate across applications intelligently. This is a fundamentally different approach.

The enterprise automation opportunity

Orby AI built a Large Action Model (LAM) platform for enterprise automation. Orby observes, learns, and automates. Users do not need to craft precise queries; the system observes workflows and suggests automations across applications. This machine-learning-enabled approach stands in stark contrast to legacy solutions, which track user-interface-based actions and force companies to hire systems integrators (SIs) and consultants to manually build each automation. Those automations are brittle and break when the interface changes.

In practice, the "observe, learn, automate" loop works like this: a finance team member processes an invoice by pulling data from an email, cross-referencing it in an enterprise resource planning (ERP) system, and logging the payment in a third application. Orby AI watches that sequence, identifies the pattern, and builds an automation that handles the entire workflow without manual scripting.

The difference from legacy RPA is structural. Traditional RPA tools record a fixed sequence of clicks and keystrokes tied to specific UI elements. When a vendor updates a dashboard layout or moves a field, the automation breaks. Orby's model-driven approach understands the intent behind the action, not just the pixel coordinates. It adapts when interfaces change.

The use cases span common enterprise bottlenecks:

• Document processing: extracting structured data from invoices, contracts, and forms across multiple formats and routing it to the correct system
• Email triage: classifying inbound messages, pulling relevant details, and triggering downstream workflows without human sorting
• Reimbursement workflows: matching expense reports against policy rules, flagging exceptions, and pushing approved claims through to payment systems

Each of these workflows touches multiple applications. That is exactly where Orby's cross-app intelligence creates the most value.

The founders behind Orby AI

Founders Bella Liu and Will Lu built Orby AI from the world of automation and AI, drawing on their prior roles at Google AI and UiPath. Their founder/market fit stood out: deep learning research experience from Google AI paired with operational knowledge of why legacy RPA deployments stall.

Bella Liu spent years at Google AI conducting deep learning research, building expertise in the neural network architectures and training methodologies that underpin Large Action Models. Will Lu brought operational depth from UiPath, where he saw firsthand where legacy RPA falls short and what customers actually need from enterprise automation.

That combination matters. Building an AI-native automation platform requires both the research rigor to develop new model architectures and the domain knowledge to ship a product that solves real enterprise problems. Bella and Will brought both.

Wing's investment and the Series A

We at Wing are proud to be day 0 supporters of Orby AI, in addition to co-leads in their $30 million Series A — a sizable round for an enterprise automation startup at this stage.

This $30 million round gives Orby the capital to scale across three priorities: expanding the go-to-market team, deepening customer support operations, and growing the product and technical teams. Congrats to Bella, Will, and the entire Orby team!

Conclusion

Enterprise automation has been a large market for over a decade, but legacy RPA approaches have plateaued as organizations hit the limits of script-based workflows. We believe LAM-based platforms like Orby represent the next wave — systems that learn from human behavior rather than requiring brittle, hand-coded scripts.

At Wing, we invest across the AI-first technology stack, and Orby occupies the autonomous applications layer, where AI models take direct action in enterprise workflows rather than simply generating text or recommendations. We see this as one of the highest-value areas for near-term investment, and we are excited to support Bella, Will, and the Orby team as they build it.

Author

Sara Choi

Sign Up for Our Newsletter