Why manufacturing automation has hit a plateau — and what will get it moving

Article

If you lead manufacturing operations or IT today, automation itself probably isn’t your constraint. In many environments, it’s working exactly as intended. Production lines are more stable. Downtime is lower. And automated systems are doing the jobs they were designed to do, often reliably and at scale.

Yet, in my conversations with plant managers, operations leaders and CIOs, a familiar theme keeps surfacing: progress feels harder than it should. Automation initiatives keep getting approved, but then momentum slows. Improvements arrive in pockets rather than end to end.

The data in Redwood Software’s new manufacturing automation research backs that up. Seven in ten manufacturers report automating 50% or less of their core operations. Only about a quarter say they’ve automated more than half.

This isn’t a failure of manufacturing automation or a lack of commitment. What the data points to instead is a structural limitation. You reach a plateau in automation maturity because automation often stops at system boundaries, not because you lack the right tools. Over time, your organization may have built an impressive collection of automation technology, but the connective tissue between those systems never quite materialized. Returns often flatten in this scenario because automation stops compounding, not because it never worked in the first place.

The middle-stage trap

When manufacturers described their automation maturity, the pattern was striking. Nearly half — 47% — placed themselves in the “Managed” stage, where automated processes exist but orchestration is partial. Another 26% identified as “Controlled,” with most tasks automated and orchestration present. Only about 2% described their operations as fully autonomous.

In other words, nearly three-quarters of manufacturers sit squarely in the middle automation maturity stages.

That clustering isn’t random. It reflects a ceiling most organizations hit after automating the obvious, self-contained processes. Early automation wins are straightforward: scheduling jobs, triggering reports, running batch processes, stabilizing equipment routines. These improvements deliver immediate value and reduce human error on the factory floor. But once those gains are captured, what remains is harder. 

The next level of improvement depends on workflows that span multiple systems — ERP, MES, supply chain platforms, quality systems and control systems built around programmable logic controllers. That requires orchestration, not just automation.

The challenge is that middle-stage maturity feels like success because dashboards are green and production rates look healthy. But the manual work hasn’t disappeared; it’s shifted into the gaps between automated processes, where people compensate with spreadsheets, emails and workarounds.

Where automation delivers and why connection matters

Automation delivers its strongest results when applied to processes contained within a single system. The report shows that about 60% of manufacturers have reduced unplanned downtime by at least 26%, with a meaningful share reporting reductions beyond 50%. Uptime, throughput and quality control consistently emerge as areas where automation excels.

These results are real, and they matter. They represent reduced risk, stabilized high-volume operations and improved consistency across production processes.

Challenges tend to emerge when outcomes depend on coordination across systems. 

• Inventory turns remain difficult to improve even as automation improves uptime, highlighting the limits of siloed execution 
• Data accuracy also lags, especially when information must move quickly between planning, execution and supply chain functions using real-time data

Lack of coordination isn’t limited to automation initiatives. Recent McKinsey research shows that broader disruptions — from supply chain volatility to shifting manufacturing footprints — are exposing the same structural weaknesses, where disconnected systems and fragmented decision-making limit performance even in otherwise well-run operations.

You can optimize maintenance schedules inside an MES or improve machining efficiency with CNC and control systems. Those are bounded workflows with clear inputs and outputs. But improving inventory performance requires synchronized data and decision-making across forecasting, production planning, material handling, warehouse operations and supplier networks.

When automation stops at system boundaries, single-system metrics improve, while cross-system outcomes lag. Orchestration addresses this gap by connecting existing automation into workflows that span the entire manufacturing environment.

The top bottlenecks between systems

When we asked manufacturers about their automation challenges, three issues arose most often: 

1. Forecasting accuracy gaps
2. Manual exception handling
3. Lack of integration between ERP, MES and PLM systems 

Together, these account for roughly 66% of reported bottlenecks. What’s notable is what isn’t on that list. Manufacturers aren’t pointing to weak automation technology, but to breakdowns between systems.

Exception handling is a clear example. Only 40% of manufacturers have automated it, even though 22% cite manual exception handling as a top disruption. Exceptions don’t respect system boundaries. A supply delay affects production schedules, inventory positions, customer commitments and financial forecasts simultaneously. Resolving that requires coordinated action across systems, not isolated scripts.

The same pattern appears in forecasting. Forecasts depend on timely, accurate data from many sources. When those systems aren’t connected through event-driven workflows, forecasts rely on stale information. By the time data is reconciled, the window for action has already closed.

These aren’t edge cases. And they persist not because automation has failed, but because automation alone was never designed to solve them.

Fragmented data automation 

Most manufacturers automate inside systems, not between them. The data shows that 78% have automated less than half of their critical data transfers. More than a quarter still move sensitive information through email or manual methods. Nearly 30% rely on scheduled scripts rather than event-driven automation that responds to conditions as they change.

Over time, this fragmentation compounds. Each new automation initiative delivers value in isolation, but also introduces another boundary that someone must manage. Complexity increases and manual handoffs multiply. Each additional project adds less incremental benefit than the one before it.

Manufacturing environments span decades of technology: legacy MES platforms, modern cloud applications, IoT and data collection layers and enterprise systems from multiple vendors. Connecting that landscape requires orchestration that can coordinate workflows across it all, based on events and business rules rather than schedules.

Reframing the challenge

Automation hasn’t failed the manufacturing industry. It has delivered real, measurable value where workflows remain contained. Fixed automation works. Flexible automation works. Individual automation solutions continue to advance.

What needs to change is the focus.

The next phase of automation maturity will be about connecting what’s already automated rather than adding more tools. Exceptions and handoffs — the points where risk and cost accumulate — need to become primary targets for improvement. Workflows must adapt in real time. How well you handle this shift will determine whether your manufacturing automation investment plateaus or continues to scale.

🠆 See a demo of what orchestration could look like using RunMyJobs by Redwood for SAP production planning.

What gets automation moving again

Manufacturers that climb beyond mid-stage maturity share common characteristics. 

• They automate exception handling across systems
• They connect data flows between ERP, MES and supply chain platforms
• They rely on event-driven workflows instead of scheduled scripts

These organizations are also more likely to explore artificial intelligence and machine learning use cases — not as a leap into the unknown, but as a natural extension of orchestrated operations. AI models are only as effective as the data feeding them, and orchestration ensures that data is timely, complete and actionable.

Orchestration changes the question from “What should we automate next?” to “Which workflows still depend on manual coordination?” It shifts success metrics from the number of automated tasks to the reduction of human intervention across the manufacturing industry.

The plateau is real, but it isn’t permanent. Changing your outcomes starts with changing how systems work together.

Get prepared for an orchestrated future now. Download the full “Manufacturing AI and automation outlook 2026” to see how your organization compares — and what it takes to move beyond the middle.

The AI and automation trends that will decide which enterprises hold up in 2026

Article

If the past few years were about proving that AI works, the next few will be about proving it can deliver.

By 2026, most enterprises will no longer be asking whether AI belongs in their automation strategy. That debate is effectively over. The harder questions are about trust, resilience and value: 

• Can automation adapt when reality does not follow the plan? 
• Can leaders rely on it when pressure is highest? 
• Does it genuinely make the business stronger, not just faster?

These questions signal a turning point. Automation is growing up. Below are Redwood Software’s top predictions for how AI, agentic systems and automation will show up in real-world IT and operations over the next year and beyond.

1. ERP will evolve from “system of record” to “system of action”

For decades, enterprise resource planning (ERP) platforms have been treated primarily as systems of record: authoritative databases and sources of truth for the business.

That’s changing. In 2026, as AI adoption expands and agentic systems move beyond chat and analysis into execution, the ERP will still be at the center of the business. But its value will increasingly come from how effectively it drives action.

This shift has been discussed for years, but only now is the surrounding ecosystem mature enough to make it practical. Many agentic initiatives struggle today because they operate in isolation, confined to a single team, department or experimental environment. They rarely deliver sustained value without deep integration into core business systems.

Service Orchestration and Automation Platforms (SOAPs) play a pivotal role in closing this gap. By connecting ERP data models via the SOAP — the orchestration layer — that span applications, integrations and infrastructure, enterprises can move from insight to execution with greater reliability. Because it allows teams to evolve processes using AI technologies with minimal disruption, a true orchestration platform enables a business’s ERP, agentic systems and traditional services to work together, making a return on AI investment far more achievable.

Watch out: Treating agentic AI as a standalone layer outside ERP and orchestration will limit its impact. The value comes when insight, decision and execution operate as one system.

2. AI governance will move from policy to operating model

Most enterprises now have some form of AI governance framework, but few have fully operationalized it. That will change quickly. 

As AI-driven and agentic decision-making becomes embedded in day-to-day operations and core automation workflows, governance can no longer live in policy decks or steering committees alone. In 2026, effective AI governance will look much more like an operating model.

This means clearly defined boundaries for autonomous action, explicit escalation paths for human oversight and transparent validation of AI models and decisions. Just as importantly, it requires auditability that scales across complex, cross-system workflows.

Strong governance is an enabler rather than a constraint, and teams move faster when they trust the systems they rely on. Organizations that build governance directly into their automation foundations will be far better positioned to scale AI responsibly and confidently.

Watch out: Governance that lives only in policy documents will slow adoption. Governance built into workflows accelerates trust and scale.

3. Shadow AI will force agentic orchestration to the forefront of enterprise operations

As AI capabilities expand, enterprises will face a familiar challenge in a new form: shadow AI.

Just as shadow IT emerged during the early days of cloud adoption, shadow AI appears when teams deploy AI tools and agents outside enterprise guardrails. These initiatives often move quickly but operate in isolation, creating fragmentation, unpredictable downtime and security exposure from tools never designed for mission-critical use.

This fragmentation is one of the main reasons many agentic initiatives stall or fail to deliver ongoing value. Intelligence without coordination means decisions are made in isolation and can’t reliably translate across complex business environments.

2026 is the year orchestration will be widely recognized as the connective tissue that resolves this problem and makes AI useful at scale. This includes the growing role of agentic orchestration, where intelligent agents coordinate decisions and actions across workflows rather than acting as standalone tools. This year, agentic AI will move from experimentation into planning. Buyers will increasingly score vendors on “agent readiness,” asking how AI agents are governed, orchestrated and integrated into existing workflows without introducing new risk.

Rather than hardcoding every possible scenario, orchestration allows workflows to adapt in real time while maintaining visibility, accountability and control. This is what turns AI from a collection of point capabilities into something enterprises can depend on.

Watch out: Shadow AI can deliver short-term wins, but without orchestration and governance, it introduces long-term operational and security risks that enterprises cannot afford.

4. AI will amplify experienced teams, not replace them

Despite the headlines, most enterprise leaders are not trying to remove people from operations. They’re trying to remove friction. This year, AI-enabled automation will increasingly support overstretched teams by handling exception triage, diagnostics and routine decision-making more consistently and at greater scale. Skilled professionals will be able to focus on higher-value work, where judgment and context matter most.

This is already changing how teams interact with SOAPs. Natural-language co-pilots are becoming standard, helping teams build workflows and configure automations without deep scripting expertise. What once required specialist knowledge is becoming accessible to a broader range of operational and technical users.

At the same time, AI-driven anomaly detection is becoming the default for runtime operations. Instead of reacting to failures, teams increasingly rely on systems that continuously ask, “What’s unusual here?” across schedules, queues, dependencies and downstream impacts — using data that orchestration platforms already collect.

This shift is critical because the IT operations skills gap is not a future problem — it’s already here. Enterprises can’t hire their way out of complexity. AI-assisted automation offers a more sustainable path by capturing expertise and making it available when and where it’s needed.

The result is better human involvement, not less. People remain accountable for strategy and outcomes, while automation absorbs the noise that slows teams down.

Watch out: AI that only accelerates development but ignores run-time operations shifts effort, not outcomes. The biggest gains come when AI supports teams across the full automation lifecycle.

➔ 40% of automation teams don’t feel ready to adopt AI. Read the latest research.

5. Resilience will matter more than efficiency

For years, automation initiatives were justified primarily through efficiency metrics: jobs automated, tickets reduced, hours saved. Those numbers were useful, until they stopped telling the full story.

By the end of 2026, enterprise leaders will care far less about how much automation is running and far more about what it protects and enables. They’ll ask:

• Did automation prevent a disruption? 
• Did it help the business absorb change without slowing down? 
• Did it keep critical commitments on track when systems, data or partners behaved unpredictably?

As enterprises become more interconnected and event-driven, resilience becomes the real measure of process maturity. Automating individual tasks is no longer enough. What matters is orchestration: the ability to manage end-to-end processes across business domains and take corrective action when conditions change.

AI will accelerate this transition by helping automation prioritize intent over rigid execution. As agentic approaches mature, automation will increasingly be able to evaluate context, choose appropriate paths and coordinate actions across systems when conditions change midstream.

Watch out: Efficiency gains from isolated automation fade quickly. Resilience comes from orchestrating processes across domains, not optimizing tasks in isolation.

What this means for 2026 and beyond

The next phase of AI and automation will not be defined by novelty, but by trust, discipline and outcomes.

It will be essential to ground intelligence in strong operational foundations, invest in orchestration and governance and use AI to empower people and focus on orchestrating work rather than automating individual tasks. As orchestration platforms take on more responsibility, enterprises can drive transformation while lowering their total cost of ownership (TCO) by reducing tool sprawl, operational friction and rework.

Automation is no longer just about doing more with less. It’s about doing what matters most, even when conditions are far from ideal.

Want help laying the foundation for agentic orchestration in 2026? Explore Redwood’s AI hub.