In November 2024 H launched Runner H, a cloud based AI agent that automates complex, repetitive tasks across industries. Unlike traditional AI models that need to be told what to do, Runner H plans and decides to achieve user-defined goals. This means it can do complex tasks without human oversight.
Key Features of Runner H
- Autonomous: Does tasks without explicit inputs, uses external tools and resources to achieve complex outcomes.
- Adaptable: Adjusts to changes in user interfaces and workflows, works consistently as environments change.
- Performance: Built on proprietary models—a Language Learning Model (LLM) and a Vision Language Model (VLM), both with 2 billion parameters—lower computational requirements.
Use Cases of Runner H
Robotic Process Automation (RPA)
Example: Data entry and form processing in financial services.
Runner H can do repetitive tasks like data entry, form processing and information retrieval. For example in financial services it can extract data from invoices and input into accounting systems, reduce manual errors and increase efficiency.
Quality Assurance (QA)
Example: Testing e-commerce websites for functionality and user experience.
In software development Runner H simplifies testing by automating repetitive testing tasks and adapting to changes in user interfaces. For example it can simulate user interactions on an e-commerce platform to ensure all features work correctly after updates, so applications are more reliable.
Business Process Outsourcing (BPO)
Example: Billing workflows in healthcare services.
Runner H automates routine tasks like billing and claims management. In healthcare it can fetch insurance plans, compile patient notes, analyze and submit claims and track reimbursements, reduce outsourcing and improve process accuracy.
Performance and Availability
Runner H has shown great performance, outperforming Anthropic’s Computer Use by 29% in the WebVoyager benchmark. Currently in private beta H has an API for developers to access pre-built agents and create custom automations, with a waitlist for interested parties.
Future Plans
H will expand Runner H beyond web environments, to universal automation where AI agents can navigate any graphical interface. This means improving accuracy, cost and developing debugging and teaching tools to tailor Runner H to specific tasks.
