Rootstock Pushes AI Agents Into Production Workflows With Summer ’26 ERP Release

Rootstock Software is moving its AI strategy deeper into day-to-day manufacturing operations with its Summer ’26 Release, introducing two native AI agents now being piloted in customer production environments alongside new security and financial management capabilities.

The release reflects a broader shift taking place across the ERP market. Enterprise software vendors are moving beyond generative AI assistants focused primarily on information retrieval and toward agents designed to participate directly in operational workflows. For manufacturers, the more consequential question is increasingly not whether an ERP platform incorporates AI, but whether that intelligence can work with live operational data and help users respond to supply, production and customer issues as they develop.

Rootstock’s latest release targets that transition with Sales and Purchasing agents built natively into its ERP platform. The agents address specific manufacturing use cases including available-to-promise and capable-to-promise analysis, lot traceability, purchase order delays and alternative sourcing.

Importantly, Rootstock says the agents are not simply part of a future product roadmap. They have entered active pilots with select customers and are already being used within production environments.

From ERP Copilots to Operational Agents

Much of the first phase of enterprise AI focused on helping users find information, summarize data or interact with business applications through natural language. Rootstock is positioning its new agents further along the operational chain.

The Sales Agent is designed to support two workflows where access to accurate, current ERP information can directly affect customer commitments.

For customer delivery promises, the agent evaluates available inventory and component requirements. If sufficient stock is not available, it can examine incoming supply to help determine when an order can realistically be fulfilled.

Its second use case focuses on lot traceability and recalls. When products from a defective or damaged lot have already been shipped, the agent can identify the affected orders and produce an action or recall list covering impacted customers.

The Purchasing Agent, meanwhile, is aimed at detecting potential supply problems before they propagate through manufacturing operations. When a purchase order due date slips, it can assess which sales and work orders may be affected and identify alternative sources of supply.

That ability to connect a procurement exception with its downstream production and customer consequences illustrates where agentic ERP could become particularly significant. Rather than requiring planners to identify a delay in one system and manually investigate its impact elsewhere, AI agents can potentially reason across interconnected ERP data and bring the operational consequences to the user.

“Manufacturers and distributors don’t have the IT bandwidth to stitch together a dozen point solutions to get AI working inside their operations,” said Rick Berger, CEO of Rootstock Software.

He argued that embedding AI within the ERP architecture allows customers to introduce agents without adding the integration complexity associated with separate AI applications.

Native AI Becomes an ERP Architecture Question

Rootstock’s emphasis on native integration also highlights an emerging competitive issue for ERP providers: where AI sits within the technology stack.

For operational AI to move beyond recommendations, agents need access to the business context contained within ERP systems — inventory positions, bills of material, purchase orders, work orders, customer demand, supply availability and financial information.

Rootstock says its agents are native to its ERP platform while also being accessible through certain workplace messaging applications and Model Context Protocol (MCP) servers. The latter could become increasingly relevant as enterprises seek to connect specialized business applications with a broader ecosystem of AI agents and interfaces.

The approach gives users the option of accessing ERP intelligence without necessarily beginning every interaction inside the ERP application’s traditional interface.

Rootstock also says the agents were co-created with manufacturing customers through its AI Council, an approach intended to focus development on practical operational requirements rather than generalized AI functionality.

“Our vision is to lead the AI-driven ERP market,” said Ohad Idan, VP of Product at Rootstock Software.

According to Idan, the agents are already helping pilot customers surface information at the point where operational decisions need to be made.

The significance of those pilots will ultimately depend on how reliably the agents perform as deployment expands. In manufacturing environments, moving AI from information retrieval into operational decision support raises the bar considerably for data accuracy, governance and user trust.

AccessGuard Targets Midmarket ERP Security

AI may headline the Summer ’26 Release, but Rootstock is also addressing another increasingly important ERP requirement: granular data access.

The company has introduced AccessGuard, its first native enterprise security solution, aimed specifically at midmarket manufacturers and distributors.

AccessGuard provides company- and division-level access controls, allowing organizations to determine which records individual users are authorized to view. The controls are embedded within Rootstock ERP rather than requiring an additional external security platform.

For manufacturers expanding across business units, divisions or geographies, these controls become increasingly important as ERP environments contain a larger volume of operational, financial and commercially sensitive information.

The addition is also relevant in the context of AI adoption. As intelligent agents gain greater access to enterprise information, identity, permissions and data boundaries become part of the AI governance discussion rather than a separate security consideration.

AccessGuard is being offered as an add-on to Rootstock ERP.

Financial Capabilities Expand for More Complex Operations

Rootstock is also continuing to strengthen its financial management capabilities for organizations operating across multiple markets.

The Summer ’26 Release introduces a redesigned Bank Statement Workbench and reconciliation experience with embedded bank connectivity, enhanced automatic reconciliation processing, broader support for European and Canadian operations, and deeper integrations with third-party accounting platforms.

These enhancements point to Rootstock’s efforts to support manufacturers whose financial requirements become more complicated as their operations expand internationally.

While manufacturing ERP differentiation has traditionally centered heavily on production, inventory and supply chain functionality, financial depth remains critical for companies seeking to consolidate more of their operations within a single ERP environment.

Customer Feedback Reaches the Product Roadmap

The release also includes Word export functionality for RootForms, allowing ERP-generated documents to be exported in DOCX format in addition to PDF.

On its own, the feature is comparatively modest alongside AI agents and financial enhancements. Its development process, however, is noteworthy: Rootstock says the functionality originated through its Idea Factory and was supported and voted on by customers.

That combination of large architectural investments and smaller user-requested improvements is becoming increasingly important for ERP vendors. AI may dominate product roadmaps, but customers continue to judge ERP platforms on the everyday usability and workflow details that affect employees across the organization.

Why This Matters for Manufacturing ERP

Rootstock’s Summer ’26 Release illustrates how the ERP industry’s AI conversation is changing.

The first wave of enterprise generative AI largely added conversational interfaces to existing applications. The emerging phase is about connecting AI with operational context and allowing agents to analyze events, trace their consequences and help initiate appropriate responses.

Manufacturing provides a demanding test case for that model. A late purchase order can affect production schedules, inventory availability, customer commitments and ultimately financial performance. A problematic lot can require rapid tracing across purchasing, production and sales records. These are precisely the situations where intelligence embedded within transactional ERP data could deliver more value than a standalone AI assistant.

At the same time, production deployment introduces questions that demonstrations cannot answer alone. Manufacturers will need evidence that AI agents can operate reliably against changing operational data, respect security and authorization boundaries, explain their conclusions and fit into existing human decision-making processes.

Rootstock’s decision to put its Sales and Purchasing agents into active customer pilots is therefore more significant than simply adding another AI feature to an ERP release.

As ERP vendors increasingly compete on agentic capabilities, the dividing line may become less about who offers an AI assistant and more about whose AI can safely understand — and eventually participate in — the operational workflows where enterprise decisions are actually made.