Most AI investments in manufacturing will not show up in the P&L.
Not because the technology doesn't work — because companies apply it to the wrong problems, measure it the wrong way, and fund it as a technology investment instead of an operating model change.
Right now, the market is flooded with AI activity. Pilots are everywhere. Dashboards are improving. Models are getting smarter.
And yet, for many manufacturers, nothing material changes. Margins don't move. Throughput doesn't improve. Working capital doesn't release.
Across mid-market manufacturers, a pattern is emerging:
- AI spend today accounts for 12% of manufacturers' IT and OT budget1
- AI spend in manufacturing is expected to scale rapidly, increasing nearly 5x by 20302
- 8–12 week payback is achievable — with discipline3
- 50–60% of transactional work can be automated4
- The real value is not cost reduction — it is capacity creation
For CFOs, the question is no longer:
"Should we invest in AI?"The difference between success and failure is not technology. It is where you apply it — and how rigorously you measure it.
Sanjay Brahmawar
QAD | Redzone
There is no shortage of ambition. There is a growing gap between AI activity — and AI impact.
Over the past year I've spent time with manufacturing leaders around the world. The conversations are consistent. There is no shortage of investment. But the gap between activity and outcome is widening.
To understand that gap, I sat down with two CFOs operating in highly complex environments — Patrick Min at Aphena Pharma Solutions, a leading North American pharmaceutical manufacturer, and Kevin Davis at AmSafe Aviation, the world leader in aviation restraint technology.
The conversation was not about technology — but about how they evaluate, measure and implement AI inside their organizations: where it works, where it fails, and what actually shows up in the P&L.
AI is not a technology initiative. It is an operational and financial discipline. And the role of the CFO is central — not as an approver of spend, but as the governor of outcomes and catalyst for change.
Most AI discussions start in the wrong place.
Companies look at a specific task and ask how technology can improve it: automate a report, speed up an approval, optimize one step, or improve an analytics layer. These are improvements. But they rarely change the business's trajectory. And in many cases, they reinforce a misconception — that AI's role is primarily in back-office efficiency.
In manufacturing, that is not where value lives.
The real opportunity sits inside core operational workflows that run through ERP and connect directly to the shop floor — planning, procurement, production, inventory, exception handling, and order execution. That is why so many AI initiatives create activity without impact.
The real issue is not identifying another use case. It is identifying the few workflows that most directly shape inventory, throughput, margin, service levels, and working capital — and then redesigning how those workflows operate inside the systems that execute them.
In tile environments, the constraint is not data. It is decision latency. Kevin Davis sees this every day. Signals come in — from demand shifts, supply disruptions, and global markets. But decisions lag. Not because the data is missing, but because systems are not designed to act on it in real time.
Analytics only matter if they reduce effort or improve decisions. Otherwise, they just create more work.Kevin Davis · CFO, AmSafe Aviation
And every hour of delay compounds:
Look across the enterprise — not just finance.
The conversation around AI in manufacturing often starts too narrowly. CFOs are asked where automation can reduce cost or accelerate reporting cycles. While those benefits exist, they rarely alter the trajectory of the business.
Patrick Min frames the issue differently. The CFO must look across the enterprise and identify where operational improvements translate into measurable financial outcomes. That requires moving beyond finance workflows and into how the business actually runs — how inventory moves, how pricing decisions are made, and how production translates into margin.
Kevin Davis sees this reality daily in a global manufacturing environment. His challenge is not visibility, but timing. Signals from China, shifts in aerospace demand, and geopolitical changes all arrive quickly — but decisions lag.
Analytics only matter if they reduce effort or improve decisions. Otherwise, they just create more work.Kevin Davis · CFO, AmSafe Aviation
AI's role is not to generate more insight. It is to compress the time between signal and action, so decisions happen before financial impact compounds.
That is where financial impact lives.
From systems of record to systems of action.
The first wave of AI in manufacturing focused on visibility: more dashboards, better analytics, faster reporting. But most manufacturing environments today are still built on systems of record — systems that capture what happened, not systems that drive what happens next.
This is where the real shift is occurring. From systems of record → to systems of action.
From insight → to execution. From latency → to responsiveness. This is not a technology upgrade. It is an operating model shift.
And it is directly tied to how CFOs think about modernization. Because the question is no longer whether existing ERP and workforce systems are valuable. They are.
The question is whether they can act in real time. Modernization, in this context, is not just replacement. It is:
- Extending existing systems with AI
- Embedding intelligence into workflows
- Connecting ERP decisions directly to frontline execution
- Enabling systems to act — not just report
This is where AI investment becomes financially meaningful. Not as a separate budget line — but to unlock the full value of existing system investments.
The highest-impact use cases are remarkably consistent.
Transactional automation in orders, procurement, and requisitions
Operational decision-making across production, inventory, and allocation
Exception handling for disruption response and corrective action
But the bigger opportunity is not in treating these as disconnected use cases.
It is in redesigning end-to-end workflows first — across ERP and the connected workforce — and then embedding AI inside them.
In practice, that means rethinking:
- how supply and production decisions are made when demand changes
- how procurement and replenishment workflows respond to signals
- how exceptions are escalated and resolved across teams
- how frontline execution connects back to planning and ERP
- how orders, approvals, and operational handoffs move without delay
Not every use case matters. You have to focus on the vital few that actually move performance.Patrick Min · CFO, Aphena Pharma Solutions
AI is not about cost reduction. It creates capacity.
One of the most persistent misconceptions about AI is that it is about cost reduction. In manufacturing, that thinking breaks down quickly.
Labor is already constrained. Reducing workload does not automatically reduce headcount. What it does is create capacity.
And critically, it allows manufacturers to extract more value from the systems and workforce they have already invested in. Kevin Davis, at AmSafe Aviation, is already seeing AI take on 50–60% of transactional workload in targeted areas. The impact is not fewer people, it is more capacity:
Without adding cost
This is operating leverage. And it is where the real financial upside sits.
We didn't remove the work — we changed who was doing it. Agents now handle roughly 60% of the process, which has freed our team to focus on sourcing and supplier strategy. That's what actually moved the needle financially.Kevin Davis · CFO, AmSafe Aviation
Patrick's perspective complements this. Growth is not simply about doing more; it is about doing the right things better. When focused on high-impact areas, AI enables organizations to scale performance without losing control.
Where discipline separates winners from everyone else.
This is the line where most companies fail. Not because they lack technology. But because they fail to align AI investment with core system performance and financial outcomes.
AI without discipline becomes:
- Scattered pilots
- Sunk cost
- Organizational fatigue
Patrick Min approaches this differently. Every AI initiative is held to a simple standard: "If it doesn't show results in 8–12 weeks, it doesn't scale."
That discipline shows up in four ways:
- Every investment is tied directly to outcomes — inventory, throughput, margin, working capital
- Investments prioritize workflows inside core systems — not isolated tools
- Payback cycles are short
- Process comes before technology
Because applying AI to disconnected tools doesn't fix the business. It must be embedded where the business actually runs.
The CFO role has changed.
AI introduces tension into the organization — move too slowly, and you fall behind — move too fast, and you waste capital. But there is a third tension emerging: How to invest in modernizing systems you already own.
ERP and workforce systems represent some of the largest capital investments manufacturers have made. CFOs are now being asked to:
- Extend them
- Upgrade them
- Or replace them
Without clear visibility into ROI. That is why the CFO now plays two roles:
Governor: Enforcing discipline. Protecting capital. Ensuring outcomes.
Catalyst: Driving modernization. Unlocking value from existing systems. Enabling change.
And balancing those roles is now essential.
AI only matters when it moves financial outcomes.
For CFOs, the clearest way to understand that impact is through execution cost — the labor and overhead required to run a function, divided by the transactions it manages. In procurement, that becomes something very tangible: the administrative cost per purchase order line.
That's where AI becomes real.
At Sonic Manufacturing, roughly 70% of order placement had already been automated through supplier APIs. The remaining 30% — unstructured emails, confirmations, and change requests — still required manual intervention. It was fragmented, inconsistent, and time-consuming. When agentic automation closed that final gap, the results were immediate.
The administrative cost per PO line dropped from ~$25 dollars to under $3 dollars. Cycle times compressed. Manual work was reduced to true exceptions. What had been an operational burden became a controlled, measurable process.
But direct cost compression within the transaction layer is only part of the story. At Farsound Aviation, the impact went further. Their expediting function dropped from four FTEs to less than one — not through elimination, but redeployment.
Those resources shifted into strategic sourcing, supplier expansion, and purchase price variance reduction — work that shows up exactly where it matters: COGS. Margin. Working capital.
This is the pattern that repeats. AI compresses work. Capacity is created. But the P&L only moves when leadership decides what happens next.
The biggest mistake is assuming automation equals savings. It doesn't — unless you deliberately reassign that capacity to higher-value work.Kevin Davis · CFO, AmSafe Aviation
Don't measure AI, measure outcomes.
CFOs don't measure AI. They measure outcomes — COGS, working capital, margin. These are lagging indicators. AI works upstream, through the levers that drive them: purchase price variance, cycle time, throughput, and manual intervention.
But the connection is not automatic. It must be designed. This is where many organizations fall short. They implement the technology — but don't close the loop.
If you can't map it to margin, working capital, or cost, it's not the right use case.Patrick Min · CFO, Aphena Pharma Solutions
AI is not a shortcut around modernization. Layering AI onto fragmented systems does not create value — it accelerates inconsistency.
Real transformation requires clean data, standardized processes, and a modern operational backbone. AI amplifies what exists — it does not fix what is broken.
If the underlying process isn't clean, automation just accelerates the problem.Patrick Min · CFO, Aphena Pharma Solutions
Structured, not chaotic. Seven steps to operational clarity.
AI does not impact the P&L directly. It impacts the leading indicators that drive it. This framework sequences the CFO's decisions — from end-game definition to selective scaling.
The final step — reallocating capacity — is where value is captured or lost. Patrick and Kevin approach this with discipline: Focused. Outcome-driven. Measurable.
This shift is not conceptual. It is operational.
ChampionAI is embedded directly into the roles that run your business — activating intelligence where decisions are made and outcomes are driven. Each Champion is purpose-built to accelerate performance across the workflows that impact throughput, margin, and working capital.
-
Procurement Champion
Eliminates administrative drag across purchasing and order management
-
Sales Champion
Accelerates revenue cycles through faster forecasting and quoting
-
Sourcing Champion
Compresses supplier cycles and improves decision quality
-
Line Lead Champion
Drives frontline productivity with real-time insight and action
-
Customer Service Champion
Speeds response times and improves order accuracy
-
Accounting Champion
Accelerates forecasting and financial planning cycles
-
Master Scheduler Champion
Reduces delays and optimizes production flow
-
Production Champion
Improves execution and speeds order completion
-
Implementation Champion
Delivers faster, lower-risk system transformation
The opportunity is not to experiment. It is to execute — with precision.
We partner directly with CFOs and operating leaders to identify where AI will deliver measurable financial impact — inside your existing ERP, operations, and frontline workflows. This is not a presentation. It is a working session designed to create clarity, alignment, and momentum.
Half-day working session
What we will do together:
- Pinpoint the workflows that drive your P&L
- Identify where ChampionAI delivers immediate value
- Align on ROI, timing and execution priorities
- Define your first step toward systems of action
Frequently Asked Questions
CFOs should measure AI through the leading indicators that drive P&L outcomes rather than measuring AI itself; specifically, purchase price variance, cycle time, throughput, manual intervention rates, and working capital release.
Unlike traditional technology investments that improve discrete tasks or reporting, AI in manufacturing must be treated as an operating model change, embedded into core ERP and shop floor workflows where it directly impacts throughput, inventory and margin.
The most common mistakes are applying AI to the wrong problems, measuring it incorrectly, funding it as a technology project rather than an operational initiative, and running scattered or bolt-on pilots that create activity without moving any financial outcomes.
AI compresses decision latency across planning, procurement, and inventory workflows, preventing the compounding effect where delayed signals cause inventory to build, throughput to slip, and margins to erode before corrective action is taken.
AI hype “promises” to magically solve complex tasks and problems without an established strategy or a need for human oversight; AI that delivers outcomes, which is embedded inside the workflows that run the business, relies on practical augmentation of routine tasks, like order execution, procurement, and production scheduling, where it compresses the time between signal and action and shows up directly in COGS and margin.
TCFOs should require vendors to map their solution directly to specific operational workflows inside existing ERP systems and demonstrate payback in a matter of weeks, not months. If a vendor cannot show measurable financial impact within an 8-12 week window, there is a high chance the investment will not scale.
Yes. Mid-market manufacturers have a structural advantage in that the highest-impact AI use cases (transactional automation, operational decision-making, and exception handling) are consistent regardless of company size. Modular deployments can usually reach measurable value in less time and without enterprise-scale budgets.
By connecting ERP decisions directly to frontline execution in real time, AI eliminates the decision latency that causes production delays. At AmSafe Aviation, agentic automation took on 50–60% of transactional workload in targeted areas, freeing capacity for throughput-driving work without adding headcount.
A credible AI business case identifies 2–4 high-impact workflows tied directly to inventory, throughput, margin, or working capital; quantifies the current cost baseline; and commits to a measurable payback within 8–12 weeks. An example of this is Sonic Manufacturing, where the administrative cost per PO line dropped from ~$25 to under $3 after agentic automation.
A CFO should reject any AI investment that cannot be mapped directly to margin, working capital, or cost improvement; relies on layering intelligence onto fragmented or unclean underlying processes; or cannot demonstrate measurable results within an 8–12 week payback window.
Grounded in practice. Two CFOs. Measurable outcomes.
Aphena Pharma Solutions is a contract pharmaceutical manufacturing and packaging company headquartered in Cookeville, Tennessee. Ranked among the top contract pharmaceutical manufacturers in the U.S., Aphena serves the prescription pharmaceutical, OTC, dietary supplement, consumer health, medical device and biologics markets from two FDA-registered facilities positioned within logistical reach of 75% of the U.S. population.
AmSafe is the world's leading provider of safety restraint products for the aerospace and defense industries, headquartered in Phoenix, Arizona with manufacturing and service facilities around the world. Virtually every commercial aircraft in the world carries AmSafe products — seatbelts, airbag restraint systems, cargo nets and cabin interior products trusted by over 600 airlines and 40 aircraft manufacturers globally.
- Cisco — State of Industrial AI, 2026.
- Research and Markets — Artificial Intelligence in Manufacturing Research Report, 2025–2030.
- Tech-Stack.com — AI Adoption in Manufacturing: Insights, ROI Benchmarks & Trends, 2025. Modular manufacturing AI deployments reach first measurable value in as little as 6–10 weeks.
- McKinsey & Company — Automation could drive up to 60% efficiency gains in manufacturing support functions, Manufacturing Asia, April 2026.