Agentic automation hits enterprise scale: Automation Anywhere, Yaskawa, and KUKA signal a new operating model for industrial AI

Automation Anywhere's Autonomous Service Desk has now fulfilled more than one billion IT service requests, auto-resolving over 80% of them on average. That single figure, disclosed by the company at its Imagine 2026 conference in Dallas and reported via PR Newswire on July 2, is the clearest proof yet that agentic AI has crossed the threshold from controlled pilot to operational backbone. The same week, Yaskawa Electric announced a robot that writes and executes its own work procedures using Google DeepMind's Gemini, and KUKA released an operating system designed to unify cobot fleets across diverse production lines. Taken together, these announcements signal a convergence: the AI agent is no longer a research concept but a production asset that operations leaders need to procure and govern now.

From pilot to production: the IT service desk benchmark

Automation Anywhere reported first-quarter fiscal 2027 results showing double-digit growth across Annual Recurring Revenue, revenue, and Current Remaining Performance Obligations, according to the PR Newswire release. The number of enterprises carrying more than $1 million in ARR grew 25% year over year; ARR from that same cohort rose 27%. That divergence matters operationally: customers aren't just renewing, they're expanding into adjacent functions.

The company's Autonomous Service Desk solution was central to two disclosed deployments. A large consulting firm used the platform to achieve an auto-resolution rate above 80%, improve productivity by more than 70%, and project nearly $2 million in cost savings, all without expanding headcount in the service desk, according to PR Newswire. Separately, Oasis Investment, part of the Al Shirawi Group, used Automation Anywhere's Autonomous Finance solution to centralize workflows across 52 companies, improving governance and auditability as transaction volumes grew.

The economics extend to the managed service provider channel. According to DeskDay research cited in the Automation Anywhere announcement, 87% of MSPs plan to increase AI investments, expecting service desk automation to cut ticket volumes by 40 to 60%. For IT procurement teams, that creates pressure to align vendor contracts with outcome-based pricing structures rather than seat licenses, a model Automation Anywhere is already executing: the company disclosed that Q1 included the single largest outcome-based transaction in its history.

Yaskawa and KUKA push agentic logic onto the factory floor

While enterprise software vendors count service requests, robotics manufacturers are embedding the same agentic logic into physical systems. Yaskawa Electric has developed a robot that integrates with Google DeepMind's Gemini to assess on-site conditions, assemble its own work procedures, and execute them independently, according to Automation International. The distinction from a conventional industrial robot is meaningful for operations planners: the system doesn't require a programmer to re-code responses to variability. It reasons about what it observes and adapts.

KUKA's contribution addresses the platform fragmentation that typically slows cobot rollouts. The company expanded its LBR iisy cobot line with the iiQKA.OS2 operating system, designed to enable flexible automation across diverse production environments, per Automation International. A unified OS across a cobot fleet reduces the integration overhead that procurement teams often discover only after purchase orders are signed, when disparate controllers and software versions turn a straightforward deployment into a months-long integration project.

Mitsubishi Electric is approaching the same challenge from a different angle. The company is combining predictive maintenance, deterministic networking, and machine vision to advance what it calls connected manufacturing, according to Automation International. The pairing of machine vision with network determinism is particularly relevant for quality-sensitive lines where latency in defect detection carries direct cost implications.

Edge intelligence fills the gap between the sensor and the cloud

Not every facility has the network infrastructure to push data to a cloud analytics platform in real time, and several hardware announcements this month target that gap directly. Igus released its i.Cee² industrial edge module, described by Automation International as a platform for data acquisition and condition monitoring built specifically to bypass the complexity of traditional digital infrastructure. The positioning is deliberate: many manufacturers have aging equipment that generates useful condition data but can't easily connect to enterprise IoT platforms without significant IT involvement.

Aetina extended its NVIDIA Jetson Thor portfolio with support for the new Jetson T3000 and T2000 modules, according to Automation International. The upcoming DeviceEdge AIE-KT and fanless AIE-PT systems are designed for robotics and industrial edge applications where power efficiency and form factor constrain what compute hardware can be installed. Power-efficient edge inference matters in environments where running full rack-mounted servers near a production line is impractical.

NEC Corporation, working with Keio University, added another layer to the edge-to-operations picture: an AI system that generates detailed 3D models in one minute from standard smartphone footage, per Automation International. The practical use case for plant operations is rapid as-built documentation and change verification without specialized scanning hardware. A maintenance or facilities team that can update a 3D site model in a minute, using equipment already in their pockets, removes a persistent bottleneck in work-order accuracy.

What this means for your team

• Revisit service desk contracts with an outcome-based lens: Automation Anywhere's 80%-plus auto-resolution benchmark and sub-eight-week deployment timeline are now documented baselines. Use them as minimum performance thresholds in RFP criteria, not aspirational targets.
• Evaluate cobot OS consolidation before the next equipment cycle: KUKA's iiQKA.OS2 and similar unified platforms reduce integration costs that typically surface post-purchase. Audit how many different robot controllers and software versions your floor currently runs before issuing the next cobot PO.
• Pilot edge modules on legacy lines before committing to full IIoT infrastructure: igus's i.Cee² and Aetina's Jetson-based systems are designed for environments where full connectivity upgrades aren't feasible. A focused edge deployment on a high-criticality asset can demonstrate condition-monitoring ROI without a facility-wide network overhaul.
• Assess agentic robot readiness with a variability audit: Yaskawa's Gemini-integrated system is most valuable where task variability is high and reprogramming costs are significant. Map which production steps currently require the most programmer intervention; those are the strongest candidates for agentic robot evaluation.