AI can help dealers move faster and see more – if they protect their data, keep people accountable and treat the ERP as the source of truth.
A delivery truck is waiting at the gate. A contractor is calling about a backorder. The yard team is trying to find a misplaced unit. Someone in the office is deciding whether to release a large order on credit. In an LBM business, the work is physical, immediate and connected. A wrong answer does not stay on a screen. It can become a missed delivery, a margin leak, a safety problem or a damaged customer relationship.
That is why the industry should be excited about artificial intelligence – and careful with it. AI can summarize long email chains, draft routine customer communications, surface unusual purchasing patterns, help employees find information and turn years of operational data into useful questions. Used well, it can reduce administrative work and give experienced people more time to solve the problems that actually require experience.
But AI is not a substitute for judgment, accountability or a sound operating process. Responsible adoption begins with a simple principle: the company remains responsible for every decision made with the help of AI. The software does not own the price, the promise, the load, the credit decision or the safety outcome. We do.
Start With the Work, Not the Technology
The best first question is not, “Where can we add AI?” It is, “Where are our people losing time, repeating work or making decisions without enough information?” That keeps the conversation grounded in a real operating need.
A practical pilot might help an inside salesperson summarize a customer thread before returning a call. It might help a purchasing manager identify items with unusual demand or suggest questions to investigate. It might help a manager create a first draft of a procedure that experienced employees then correct. Those uses assist people without silently taking control of the business.
Each use case should have a named owner, a clear purpose and a definition of success. If the goal is to reduce order-entry rework, measure rework. If the goal is to answer customers faster, measure response time and accuracy. Simply adopting AI does not demonstrate success; success means a measurable improvement in how the business operates.
Keep the ERP as the System of Record
AI should work alongside the systems and processes a company already depends on. It should not become a shadow system where prices, inventory quantities, credit terms, purchase orders or delivery commitments drift away from the ERP.
An AI assistant may help a user find or interpret information, but the authoritative transaction should still be created, approved and recorded through the proper business system. If AI recommends a reorder quantity, an authorized person should review the underlying demand, lead time and commitments before placing the order. If it drafts a customer response, the employee should verify availability, price and delivery timing before sending it.
This distinction matters. AI is very good at producing a confident answer. Confidence is not the same thing as correctness, and a polished sentence is not a transaction record.
Protect the Information That Makes the Business Valuable
LBM companies hold sensitive information: customer pricing, vendor costs, rebate structures, credit details, employee records, takeoffs, plans, jobsite information and years of purchasing history. Employees should never assume it is safe to paste that information into a public AI tool.
Before approving a tool, management should understand what data it receives, where that data is stored, who can access it, whether it is used to train a model, how long it is retained and how it can be deleted. Access should follow the same role-based principles used elsewhere in the company. A salesperson does not need unrestricted access to payroll data, and an AI assistant working for that salesperson should not have it either.
The rule for employees should be easy to remember: if you would not send the information to an unknown outside party, do not enter it into an unapproved AI service.
Put People at the High-Consequence Decision Points
Not every AI use requires the same level of control. Drafting a meeting agenda is low risk. Changing a price, extending credit, selecting a job candidate, issuing safety instructions or changing the sequence of a load can carry real financial, legal or physical consequences.
The higher the consequence, the stronger the review should be. Companies should define which actions AI may assist, which require a specific human approval and which are off-limits. In practice, that means no autonomous changes to pricing, credit, payroll, hiring, safety procedures, purchase orders, inventory adjustments or delivery commitments without an authorized person reviewing the source information and taking responsibility for the decision.
Human review cannot be ceremonial. The reviewer needs enough context, time and authority to challenge the recommendation. Clicking “approve” on an AI recommendation without checking the information behind it is not meaningful oversight.
Verify Outputs and Make Mistakes Visible
AI can fabricate facts, omit context and repeat bias found in its data. The answer may sound certain even when it is wrong. Every company using AI needs a verification habit: compare important outputs with source documents, ERP records, approved policies and the knowledge of the person accountable for the work.
It also needs a way to report problems. Employees should know how to flag a bad output, a privacy concern or an unexpected action without being blamed for slowing down adoption. Managers should track those incidents, look for patterns and decide whether the tool, the data, the process or the training needs to change.
The National Institute of Standards and Technology describes AI risk management as a continuous cycle of governing, mapping, measuring and managing. That is a useful mindset for an LBM company: set the rules, understand the use, test the result and improve the controls.
Choose Vendors Carefully and Train the Whole Team
A vendor should be able to explain its security practices, data terms, access controls, logging, model limitations and incident process in plain language. If those answers are vague, the risk has not disappeared; it has merely been transferred to the customer.
At the same time, a policy sitting in a binder will not protect the company. Employees need short, practical training built around situations they actually face: Can I upload this takeoff? May I use AI to write a performance review? Who checks a suggested price? What do I do when an answer conflicts with the ERP? When must I tell a customer that content was generated or materially shaped by AI?
Leaders should follow the same rules they expect employees to follow. Responsible use becomes part of the culture only when management treats security, accuracy and accountability as operating disciplines, not IT fine print.
Move Deliberately—but Keep Moving
Responsible does not mean afraid. It means starting with a useful, contained problem; using approved data; keeping a person accountable; measuring the result; and expanding only after the process earns trust.
LBM has always adopted technology in the context of real work. Barcodes, mobile devices, ERP systems and fleet tools created value when they fit the operation and when people understood how to use them. AI should be held to the same standard.
The companies that benefit most will not be the ones that chase every new tool. They will be the ones that combine modern capability with the habits that already make a strong dealer: accurate information, clear responsibility, trained people, disciplined processes and promises that can be kept.
Kenny Mays is the founder of LMBR Technologies, an AI-native operating system built specifically for the LBM industry and is designed around the people who do the work and the operational realities they face every day. LMBR supports lumberyards, truss plants, roofing distributors, drywall supply, and hardscape suppliers. For more information, visit trylbmr.com or email Kenny at kenneth@trylmbr.com.
