AI for Manufacturers: It's Not Just the Factory Floor
Sajana Weerawardhena, Co-founder, Preconfigured
· Updated
AI for manufacturers quotes a configured product, triages a warranty claim, answers a service case or predicts a failure, working off the manufacturer's own data in CRM, ERP and service systems. Most guides start on the factory floor. The faster payback usually sits in sales and service, where that data already lives and the work is repetitive. In production, this has cut dealer quote turnaround from a week to a day and order-change errors by 62%.
The Adoption Gap
Adoption numbers for manufacturing AI disagree. A 2026 survey of 1,200 manufacturing leaders puts adoption at 72%. The same survey finds only 10% have scaled AI across their networks. SYSPRO research with Frost & Sullivan finds fewer than 20% of complex manufacturers have deployed AI or are implementing it.
The spread is a definition problem. “Using AI” counts a chatbot pilot; “scaled” means running in production across sites. Only the second number describes value — the value is not in chatbots, it is in agents that do real work across the systems of the enterprise.
The Factory-Floor Bias
Most published advice covers four use cases. Predictive maintenance. Vision inspection. Demand forecasting. Supply chain planning. Each is real. Each also depends on sensors, historian data and plant-by-plant rollout.
The commercial side gets far less attention. It leaks money every day. In Aleran's survey of 200 B2B manufacturers, manual sales and quoting cost an average of 5% of revenue a year. 88% reported lost deals. Salesforce research finds 78% of manufacturers spend substantial time searching for information across systems.
The Opportunity in Sales, Service and Warranty
Sales and service lose revenue and margin in plain sight. Five problems recur across manufacturers.
| Problem | How it shows up | P&L line |
|---|---|---|
| Slow quotes | Days to quote a configured product; the buyer goes to the faster supplier | Revenue, win rate |
| Quote and order errors | Invalid configurations, wrong prices, rework after the order lands | Margin, cost to serve |
| Discount leakage | Reps discount to close without seeing margin impact | Gross margin |
| Slow service response | Cases wait while staff hunt for asset and order history | Retention, cost to serve |
| Warranty leakage | Out-of-entitlement claims paid; supplier recovery missed | Warranty cost |
Each problem maps to a line the CFO already tracks. The baseline exists in today's reports. The return needs no new instrumentation to measure.
The Cross-System Problem
Quoting and service work rarely lives in one system. A quote for a configured product touches CRM for the account, CPQ for the configuration, ERP for price and lead time, engineering for build rules, and email for the back-and-forth. A warranty claim touches the service system, ERP serial and shipment records, supplier terms and quality data.
People are the integration layer. A rep or coordinator carries context from screen to screen, rekeys data and waits on other teams. The delay and the errors sit in those hand-offs.
Most AI pilots run inside one system. A CRM copilot summarizes the account. An ERP assistant answers a stock question. Each works. Neither touches the hand-offs.
Proof From Production
Three recent Preconfigured deployments show the pattern. Each workflow crossed four or more systems. Each reached production in under two months.
| Manufacturer | Workflow | Systems crossed | Live in | Result |
|---|---|---|---|---|
| $2B+ automotive parts | Dealer RFQ quoting | Email, dealer spreadsheets, product catalog, ERP | 7 weeks | Quote turnaround cut from up to a week to one day; 90%+ of lines resolved automatically |
| $5B agricultural inputs | Customer order changes | Email and cases, Salesforce, ERP, warehouse, transportation | 6 weeks | Changes in minutes instead of hours; 62% fewer errors and avoidable redeliveries |
| $7B+ power equipment | Aftermarket opportunity from technician notes | SAP, Salesforce Field Service, Microsoft Teams | 4 weeks | $15M in new aftermarket opportunity in one division |
The auto parts bottleneck was finding parts, not pricing them. Reps worked dealer spreadsheets line by line against the catalog and the ERP. A 300-line request now arrives with about 285 lines already resolved.
The agricultural team sat in the middle of every order change. One email could touch the customer, the ERP, the warehouse, the carrier and sales. The team now supervises changes instead of coordinating them.
The power equipment division ran sales and service as separate worlds. Contracts lived in SAP. Technician knowledge lived in Field Service notes. Uncovered assets now reach sellers as opportunities inside Teams.
Frequently asked questions
- What is AI for manufacturers?
- It is software that uses a manufacturer's own data to make or recommend decisions. Common uses are quoting, service, warranty, maintenance, quality inspection and forecasting.
- Where should a manufacturer start with AI?
- Start with one sales or service problem that has a measurable cost, such as slow quotes or warranty leakage. The data already exists in CRM and ERP, so no sensors are needed.
- Why do manufacturing AI pilots stall?
- Most automate work inside one system. The cost sits in work that crosses CRM, CPQ, ERP and email, which a single-system pilot never reaches.
- What results have manufacturers seen from AI in sales and service?
- An automotive parts manufacturer cut dealer quote turnaround from up to a week to one day. An agricultural manufacturer cut order-change errors by 62%. A power equipment maker found $15M in aftermarket opportunity in one division.
Sources
Sajana Weerawardhena is co-founder of Preconfigured. Previously led the machine learning team at Robust Intelligence, acquired by Cisco. More about the team.