Manufacturing is in the middle of a once-in-a-generation reset. Tariffs, reshoring and legislation like the CHIPS and Science Act are restructuring supply chains and driving billions in new investment. 

Across the U.S., new plants are being built. But the pressure is just as real as the opportunity. Volatile demand, persistent labor shortages and a global market that favors whoever moves fastest.

In a recent Gartner piece, QAD | Redzone CEO Sanjay Brahmawar explored why this moment demands a new kind of ERP. Gartner’s Greg Leiter and Denis Torii echoed similar sentiments in the research note, Innovation Insight: AI Is on the Cusp of Reshaping ERP. Combined, the articles are a powerful indicator that the systems manufacturers have long relied on are fundamentally unprepared for the complexity they now face. 

The solution isn’t a better version of the same tool. It’s a new operating model entirely.

The Problem with Traditional ERP for Modern Manufacturing

ERP was never designed to make decisions; it was designed to track them. A transaction happens, the system records it. An inventory level drops, someone notices and reacts. A production schedule changes, someone rebuilds it manually. Every step involves a person in the loop, which was fine when manufacturing moved at a certain speed. It’s less fine now.

“Modern, agentic ERPs don’t just preserve history—they proactively influence future outcomes by initiating and executing next-best actions.” shared Sanjay.

The problem isn’t a lack of investment or effort. It’s simply that legacy ERP systems, especially those still running on-premise, weren’t built to sense changes in real time or reason through options. The result is slower decision-making, reliance on manual processes and a disconnect between the speed of the market and the speed of the business.

What Makes Agentic AI Different?

Agentic AI consists of autonomous software agents that perceive context, learn from patterns, make real-time decisions, and execute tasks with human oversight—escalating when guardrails require.

Humans define goals, constraints, and guardrails; agents execute within these boundaries and escalate exceptions for review, ensuring accountability and control.

And that’s significant, explained Sanjay, because agentic AI is so much more than a layer of intelligence. It’s an operating model for how ERP works. Embedded AI agents can detect anomalies, simulate outcomes, recommend actions and even execute next-best steps autonomously, escalating only when needed.

The Value of Agentic AI on the Factory Floor

The power of agentic AI is visible in several critical aspects of manufacturing operations. Supply chain management is one of the most immediate. AI agents can identify unusual supplier transactions and manage the initial resolution steps independently, empowering procurement leaders to focus on other important tasks.

Agentic AI is also a game changer on the labor side, which is a massive opportunity considering the industry is facing a 3.8 million skills gap by 2033. It can help with simple tasks like automating job posts and candidate sourcing, but also address a much larger challenge: preparing the next generation of manufacturing workers. Younger workers demand technology that feels as intuitive as the apps and devices they interface with every day, making outdated systems a recruiting liability. Modern, conversational AI interfaces are essential.

On the plant floor, agentic AI can model multiple production scenarios, create new schedules in response to shifting demand and surface actionable insights from data that would otherwise sit in a dashboard waiting for human review. In an environment where keeping up with delivery demands and shifting customer expectations are key, agentic AI provides the competitive edge that modern manufacturers need. 

For new facilities, the technology is especially transformational. Manufacturers using QAD’s Agentic AI capabilities have gone live in 90 days. Traditional ERP implementations at greenfield sites have routinely taken six to eighteen months.

Cross-System Agent Scenarios

The real power of agentic AI emerges when workflows span systems and teams. Instead of operating in silos, agents coordinate across ERP, MES, SRM, planning, and connected workforce systems to respond to events in real time — at the seams where execution typically breaks down.

Scenario 1: Demand change triggers coordinated response
A customer demand change exceeds forecast. An agent:

• checks ERP for available inventory and capacity
• evaluates the change against supplier commitments and cost constraints
• proposes adjusted production schedules and material actions for approval
• updates delivery commitments and customer order status on approval

Impact: Improved OTD, reduced manual replanning, faster response to demand shifts

Scenario 2: Supplier disruption triggers supply chain replan
A late or constrained delivery signal is detected in SRM. An agent:

• identifies affected orders and production lines in ERP
• evaluates tradeoffs across alternate suppliers, inventory buffers, cost, and service
• recommends rebalanced production schedules for review
• alerts customer-facing teams with updated timelines once actions are confirmed

Impact: Reduced downtime, better schedule adherence, proactive customer communication

Scenario 3: Production exception triggers cross-functional coordination
A quality hold, scrap event, or line disruption is flagged on the shop floor. An agent:

• assesses the impact on schedules, inventory usage, and delivery dates in ERP and MES
• coordinates response across operations, quality, and planning
• proposes adjusted schedules and inventory actions for approval
• surfaces updated delivery commitments to customer service

Impact: Faster issue resolution, maintained throughput, minimized disruption to delivery

Gartner’s Hot Take on Agentic AI in ERP

The market data makes the future of agentic AI clear. In 2024, 14% of ERP spending went to applications with GenAI or agentic AI capabilities. Gartner forecasts that number to reach 62% by 2027. That’s not gradual adoption – it’s a market undergoing massive change in a three-year window. 

Gartner’s Greg Leiter and Denis Torii also had some good advice for manufacturers considering agentic AI: keep an eye out for “agent-washing” – the practice of relabeling chatbots or basic automation as AI agents. Instead, anchor every vendor evaluation to a specific, proven use case. And don’t forget that quality is king. Agentic AI with weak data governance doesn’t just underperform. It can compound bad decisions faster than a human would. 

For most manufacturers, the practical question is where to start. Gartner recommends identifying the specific processes that cost the most in delay or manual effort, then mapping them to available AI use cases with proven outcomes. 

Agentic AI in ERP, like ChampionAI, is no longer theoretical. The competitive gap between manufacturers using traditional ERP and intelligent systems like Adaptive ERP will continue to widen. The real question is whether to move now or wait for competitors to make the move first.

How Agentic AI Integrates with Your ERP

To move from theory to execution, manufacturers need to understand how agentic AI actually connects into existing ERP environments. This isn’t a rip-and-replace model. For example, ChampionAI operates as a Manufacturing Agentic Layer that sits across your data, workflows, and execution systems to turn insight into action, coordinating work at the seams between teams and systems, where execution typically breaks down.

At a high level, ChampionAI operates as a coordinated architecture across six core layers:

• Data services and APIs
  Agentic AI connects to ERP and adjacent systems (MES, SRM, Supply Chain, Connected Workforce) through structured APIs and data services. This includes:

• • transactional ERP data (orders, inventory, suppliers, production schedules)
  • master data (products, BOMs, pricing, vendors)
  • real-time operational signals from plant systems

This layer ensures the AI is working from the same system of truth as the business.

• Event triggers and signals
  Rather than relying on static reports, agentic AI listens for events across the enterprise. These can include:

• • supply chain disruptions
  • demand spikes or forecast variance
  • production delays or downtime
  • labor gaps or scheduling conflicts

Events act as the “starting gun” for agent-driven workflows, enabling real-time responsiveness instead of delayed human intervention.

• Orchestration and reasoning layer
  Unlike generic automation platforms, this layer reasons with manufacturing-specific context, understanding constraints, policies, and economic tradeoffs that horizontal tools can’t. The orchestration layer:

• • interprets events in context of business goals and operational constraints (capacity, inventory, supplier commitments, cost, margin)
  • runs scenario analysis and simulations
  • determines the optimal course of action
  • coordinates execution across multiple systems, teams and functions

Instead of isolated automation, this layer enables cross-functional decision-making at machine speed.

• Action execution and write-backs
  Once a decision is made, agentic AI doesn’t stop at recommendations. With the appropriate approvals in place, it can:

• • update ERP records (e.g., adjust orders, reschedule production)
  • trigger downstream workflows (procurement, logistics, staffing)
  • initiate communications with suppliers or internal teams

All actions are written back into the ERP, ensuring continuity, auditability, and system integrity.

• Outcome attribution

ChampionAI closes the loop by connecting decisions to measurable results: identifying what changed, why, and which agent drove it. This provides verifiable ROI, strengthens trust in autonomous decisions over time, and ensures improvements show up where they matter most: in the P&L.

• Human-in-the-loop controls
  Autonomy doesn’t mean absence of control. Manufacturers can define thresholds and guardrails such as:

• • approval requirements for high-impact decisions
  • escalation paths for exceptions
  • visibility into agent decisions and actions

This ensures trust, compliance, and alignment with operational policies.

Together, these layers turn your system of record into a system of action, reducing the manual coordination that drives cost and delay, and delivering measurable improvements in cost, productivity, and service without a multi-year ERP project.

With agentic capabilities, ERP evolves from a passive system of record into a proactive system of action, an intelligent partner that makes and executes decisions to navigate complexity and achieve strategic goals.

A 6-Step Playbook to Operationalize Agentic AI

Moving from pilots to production doesn’t have to mean a multi-year systems overhaul. The manufacturers that scale successfully start in targeted workflows, prove measurable impact, and expand in a structured way. The key is aligning workflows, value, and ownership from the start.

1. Target high-impact workflows
   Prioritize the cross-functional, exception-heavy processes where work crosses teams and systems — procurement, sourcing, sales, production scheduling — and where people spend their day coordinating through email, spreadsheets, and meetings. These seams are where agentic AI delivers the fastest impact.
   KPIs: OTD, schedule adherence, expedite frequency, purchasing cost
   RACI: Operations lead; finance and IT support
2. Define value upfront
   Set clear success metrics tied to business outcomes before deployment, so impact shows up in the P&L. Build the financial and operational case — ROI, total cost of ownership, and opportunity mapping — before committing to scale.
   KPIs: cost of goods, working capital, cycle time, margin
   RACI: Finance defines value; operations owns delivery
3. Start where you are
   The most successful programs meet the business where it is, with flexible entry points and no required system migration. Deploy agents into real workflows quickly, rather than waiting for a large transformation to complete before seeing value.
   RACI: Business sponsors; operations and IT align on entry point
4. Connect data and workflows
   Agents should work from the same system of truth as the business, integrating with ERP and adjacent operational systems without replacing them. Agents that reason with manufacturing-specific context — constraints, policies, and economic tradeoffs — make accurate decisions from day one, while the ERP remains the governed system of record.
   RACI: IT and data governance own integration; business validates
5. Deploy with guardrails
   A semi-automated model builds trust: routine tasks run autonomously, while items requiring judgment are escalated for approval before an agent executes, updates records, or communicates externally. Define the thresholds, escalation paths, and visibility that fit your risk tolerance.
   KPIs: exception rate, override frequency, decision speed
   RACI: Operations defines policy; IT implements
6. Scale and prove impact
   Expand into adjacent workflows and functions, and use closed-loop outcome attribution to connect each decision to measurable results, verifying ROI and strengthening trust as you scale. Improvements compound across the business without adding complexity or headcount.
   KPIs: working capital, service levels, productivity, margin
   RACI: Executive leadership sponsors; cross-functional teams execute

How Will This Work with My Legacy ERP?

Manufacturers don’t need to replace existing ERP systems to adopt agentic AI. Most deployments follow a coexistence model that layers intelligence on top of current environments.

Can agentic AI work with on-prem or legacy ERP systems?
Yes. Agentic AI connects through adapters, APIs, and integration layers that expose ERP data and workflows without requiring full system replacement. This allows existing investments to remain in place while adding new capabilities.

How does integration typically work?
Most architectures use an integration layer to capture and act on signals across systems. But the difference between generic automation and agentic AI is what happens in between: rather than simply routing messages, the orchestration layer reasons with manufacturing-specific context — constraints, policies, and economic tradeoffs — to decide what should happen next. In practice, it: 

• captures signals from ERP and adjacent systems
• interprets them against operational goals and constraints 
• triggers and coordinates agent-driven workflows in real time 
• write updates back into the ERP

This creates a hybrid model where legacy systems remain the system of record, while agentic AI becomes the system of action.

Do we need to modernize all data upfront?
No. Manufacturers typically take a hybrid data approach:

• prioritize critical workflows and datasets first
• standardize key master data over time
• incrementally improve data quality as agents scale

This avoids long, disruptive transformation cycles.

What does a phased rollout look like?
Most organizations start with a small number of high-value processes, such as:

• supply chain exception management
• production scheduling
• procurement workflows

From there, they expand into adjacent areas, reusing integration patterns and orchestration logic.

What are typical timelines?
Initial use cases can be deployed in as little as 90 days, with broader scaling occurring over several months as additional workflows are onboarded.