LittleHorse adds AI agent tools & free serverless trial

LittleHorse has released version 1.3 of its Saddle Command Centre software.

The update adds new tools for creating AI agents and introduces a free serverless trial option.

The release is aimed at organisations building AI applications that need to connect business processes with software workflows. New features in version 1.3 include pre-built task workers, agent skills and a JavaScript software development kit for developers.

LittleHorse describes Saddle Command Centre as an action layer between business requirements and code. The software is designed to help organisations orchestrate agents, microservices and events across distributed systems.

The company was founded in 2022 by Colt McNealy, who previously worked on software integration problems across sectors including mortgage, insurance, scheduling and operations. According to LittleHorse, that experience led to the view that businesses faced orchestration issues before the arrival of AI agents, and that newer AI systems have made those issues more visible.

Scott McNealy, Co-Founder and former Chief Executive Officer of Sun Microsystems, backed the company and now serves as an adviser. LittleHorse says it has been in production with enterprise customers for more than 18 months.

Developer focus

The new serverless trial is intended to lower the barrier for teams that want to test the software without setting up their own infrastructure. The platform already supports Java, Python, Go and C#, and the latest release adds a JavaScript SDK.

In a statement on the release, Colt McNealy outlined the company's rationale for the new offering.

"We wanted to make it easy to experience the power of LittleHorse, and that is why we are so excited about the serverless offering. We want more developers to build properly guard-railed AI apps using Business-as-Code, and the serverless offering makes that possible. The release of the JavaScript WfSpec SDK is also motivated by our developer-first ethos," said Colt McNealy, Founder and Chief Executive Officer of LittleHorse.

The company describes Business-as-Code as an approach for translating business processes into executable workflows while keeping those workflows aligned with operational requirements. It argues that this model can help companies manage AI agents within governed and observable processes rather than through ad hoc integrations.

Customer use

LittleHorse cited Sejal Learning Systems as an example of how the software is being used in an AI-based learning delivery setup.

"As a service provider operating online learning academies, we needed a platform to enable integrating AI into our workflows. LittleHorse was the missing piece for us to implement and orchestrate stateless AI workers into the architecture," said Leo Da Cunha, Chief Learning Systems Architect at Sejal Learning Systems.

"LittleHorse is officially an embedded and critical component of the Sejal Learning Solutions AI-Powered Learning Delivery System and is delivering 3x-4x speed increases without adding any new hardware. We are now planning to roll this out to more of our learning academies," Da Cunha said.

Users of the platform have reported faster deployment of new services, lower spending on software-as-a-service tools and less time spent tracing failures in complex systems, according to the company. It added that the platform can also be used to integrate real-time data flows across different parts of an organisation.

Market view

The release comes as companies continue to search for practical ways to move AI agents from pilot projects into day-to-day operations. For many businesses, a key issue is not just model performance but how those systems connect with existing workflows, controls and software estates.

Brad Shimmin of Futurum said the gap between interest in agents and business results often comes down to fragmented processes and disconnected systems.

"Although frontier models are now capable of performing complex tasks, few enterprises have succeeded in transforming their operations and workflows with agents. The gap between potential and realized value is due mostly to the fragmentation of processes across tribal knowledge, sprawling SaaS systems, and unwieldy glue code," said Brad Shimmin, VP and Practise Lead, Data and Analytics, Futurum.

"Business-as-Code is an innovative approach to the last-mile delivery problem of agents because it gives agents freedom to perform creative and fuzzy tasks without sacrificing determinism, observability, durability, and compliance for the rest of these processes," Shimmin said.

Version 1.3 is available now alongside the company's open-source kernel and the new serverless trial environment.