Automation Anywhere claims 30% Agent accuracy lift
Automation Anywhere has unveiled new platform capabilities for orchestrating AI agents, automations and human workers within a single governed enterprise process.
The release targets organisations moving AI from pilots into production operations. CEO Mihir Shukla said the autonomous enterprise requires a system that coordinates how work runs within departments and across the organisation.
New universal orchestration capabilities sequence tasks across departments and platforms including Salesforce, ServiceNow and SAP. The vendor said the orchestration layer handles approvals, data sources, APIs, automations, AI agents and human decisions in one process.
Automation Anywhere also introduced Automation Anywhere Code (AAI Code), a low-code tool for building enterprise applications. The vendor said AAI Code can produce apps complete with UI, processes, agents, context and security in as little as one week.
The company described AAI Code as distinct from vibe coding tools because it plans before building and includes governance controls before deployment.
A new Context Intelligence Graph sits within the Process Reasoning Engine. It is designed to surface relevant context for each task rather than exposing all enterprise data to every step.
Automation Anywhere said agents using the engine demonstrated more than 30 per cent higher accuracy in internal evaluations across multiple business domains.
For governance and risk teams, the additions include AI Evaluations to assess agent performance at design and runtime. A new Process Simulation, Optimization and Testing environment stress-tests entire processes including exceptions and edge cases before deployment.
The vendor cited University Hospitals of Leicester NHS Trust as targeting 50 to 70 per cent autonomy for administrative work. Anticipated benefits include a 22-day reduction in recruitment time and a one million pound annual cut in temporary staffing costs.
These figures are described by the vendor as anticipated benefits and have not been independently audited.
Availability is staggered. AI Evaluations is available now. AAI Code and EnterpriseClaw are in public preview, with general availability later in 2026. Context Intelligence Graph is in preview with general availability scheduled for Q3 2026.
Process Simulation, Optimization and Testing is scheduled for preview in Q3 and general availability in Q4 2026.
https://www.automationanywhere.com
