Five Developer-First Features Driving Agentic Automation in .38

Automation Anywhere’s latest .38 release is here, and it delivers meaningful upgrades for developers building resilient, scalable, and AI-driven automations. From cloud-native infrastructure to agentic orchestration, this release focuses on helping teams move faster while handling real-world complexity with confidence.

Let’s break down the top 5 developer-focused features, starting from number 5 and working our way up to the innovation bringing a ton of excitement in this release.
 

#5 – Enterprise-Grade Automation Resiliency with Interrupt Handler

Automation resiliency takes a big step forward in the .38 release with enhancements designed to keep automations running smoothly when reality intervenes. The standout addition is the new Interrupt Handler package.

Interrupt Handler allows automations to recover gracefully from unexpected events such as pop-ups, alerts, or re-authentication prompts that would normally pause or break execution. When an interruption occurs, the automation follows pre-defined recovery logic and continues from where it left off, without manual intervention.

By reducing failures caused by transient UI or system events, Interrupt Handler significantly improves reliability and cuts down on rework. For developers managing large-scale enterprise automations, this translates to fewer escalations, less downtime, and up to 80% more resilient processes.

#4 – Automation Cloud Service Now Available on Private Cloud

With .38, Automation Cloud Service is now available on Private Cloud, giving organizations full control over security and infrastructure while still benefiting from real-time, auto-scaling automation.

Built on a Linux-based, containerized architecture, the platform eliminates the need for Windows virtual machines, reducing infrastructure complexity and cost. Developers gain millisecond-level performance and low-latency execution, making it well-suited for high-speed, enterprise-grade use cases.

The service scales horizontally to support millions of AI executions, while more than 100 pre-built packages and connectors make it easy to integrate with enterprise applications and large language models. The result is a modern, cloud-native foundation for building and running automations at scale behind your own firewall.

#3 – Agentic Orchestration Inside the Mozart Orchestrator

AI agents are now fully embedded into Mozart Orchestrator (formerly Process Composer), enabling developers to visually orchestrate AI Agents, Task Bots, API Tasks, Forms, and human interactions within a single environment.

This unified orchestration layer supports both deterministic workflows, where every step is explicitly defined, and autonomous solutions, where AI Agents determine the best path forward based on the current state.

Agents can be equipped with tools such as Cloud Service Tasks, Task Bots, Forms, and human-in-the-loop approvals, making it easier to design intelligent processes that adapt dynamically while staying governed and observable.

#2 – High-Scale Cloud Extraction Service for Document Processing

The High-Scale Cloud Extraction Service in .38 rethinks how high-volume document processing is built and scaled. Designed for cloud-native execution, data extraction runs entirely in the cloud without Bot Runners, additional licenses, or infrastructure management.

The service scales instantly to handle millions of documents with no performance degradation, delivering low-latency, high-speed extraction suitable for real-time and front-office scenarios. Developers can expect up to 5 times faster processing and more than a 20% cost savings compared to traditional Document Automation setups.

Getting started is straightforward: simply select the Cloud Extraction option when creating a new learning instance, and the service handles the rest.

#1 – Autonomous AI Agents for Goal-Based Enterprise Automation

One of the most impactful updates in the .38 release is the introduction of Autonomous AI Agents, which moves enterprise automation beyond rule-based execution to goal-driven systems.

Instead of following fixed scripts, these agents work toward defined business objectives. They plan, act, and adapt based on context, selecting the right tools for each situation. This flexibility allows automations to respond dynamically rather than breaking when conditions change.

Human-in-the-loop collaboration is built-in by design. Agents know when to pause, request clarification, or escalate decisions to humans, ensuring accuracy and alignment with business intent while still reducing manual effort.

To power these agents, developers can use Automation Anywhere–hosted LLM models, available directly within the Control Room. These models are secure, governed, and purpose-built for agentic workloads, enabling AI agents to reason, act, and operate safely across enterprise environments without external setup. Your data stays private and is not used for LLM training purposes.

Bonus: Expanded Package Ecosystem

Alongside the core features, .38 introduces several new and enhanced packages. New additions for Microsoft Teams, Microsoft Word, and Interrupt Handler improve collaboration and resiliency, while existing packages such as Mail Iterator, Apple Numbers, and Cloud Service Tasks have been enhanced for performance and flexibility. Support for Apple Keynote is also now available for creating and editing presentations.

These updates extend automation reach across more applications and use cases while improving overall reliability.

Final Thoughts: Which Feature Stands Out to You?

With the .38 release, Automation Anywhere continues to evolve toward cloud-native, agentic, resilient automations. Whether you’re focused on scalable infrastructure, intelligent orchestration, or goal-based AI Agents, this release delivers practical tools to build smarter automations faster.

Visit the Automation Anywhere University portal for training and hands-on exercises to start exploring these features.

Now it’s your turn—which .38 feature are you most excited to try? Share your thoughts in the comments below.