A sales rep at a steel service center logs into the phone system on a Monday morning. Of the first 15 calls, 4 are customers asking "do you have X in stock," 3 are asking "where is my order," 2 are requesting MTRs for shipments received last week, 2 are asking for pricing on standard products, and 4 are actual selling conversations about new orders, project quotes, or complex specifications. Only 27% of the calls involve actual selling. The rest are information retrieval that a well-built system could handle automatically.
What AI Can Handle Today
Modern AI chatbots integrated with your inventory and order management systems can answer "do you have X in stock" questions instantly by querying real-time inventory data, provide order status updates by pulling tracking information from your shipping system, deliver MTRs and quality documentation by retrieving documents linked to specific orders or heat numbers, quote standard products where pricing follows defined rules (base price plus markup by customer tier), and answer common questions about your capabilities, delivery areas, and policies.
These are not hypothetical capabilities. Companies across B2B distribution are deploying AI assistants that handle 40% to 60% of routine customer inquiries without human involvement. The technology has matured to the point where a well-configured AI chatbot provides accurate, helpful responses that customers prefer to waiting on hold.
What AI Cannot Handle
Complex quoting involving multiple products, custom processing, and project-specific requirements still needs a human. Negotiation on pricing, terms, and delivery schedules requires judgment and relationship skills that AI does not possess. Quality disputes and complaint resolution demand empathy, investigation, and authority to make things right. Technical recommendations (which grade to use, which coating to specify, which processing method is appropriate) require deep product knowledge and understanding of the customer's application.
The goal is not to replace your sales team. It is to eliminate the 70% of their time spent on information retrieval so they can focus on the 30% that actually generates revenue and builds relationships.
Implementation Approach
Start with a customer-facing chatbot on your website that handles the three most common questions: inventory availability, order status, and document requests. These require integration with your ERP or inventory system, your order management system, and your document management system.
The integration is the hard part. An AI chatbot that cannot access real data is useless, it will make things up or give generic responses that frustrate customers. The chatbot must query your actual inventory database, your actual order tracking system, and your actual document repository. This requires API connections between the chatbot platform and your business systems.
Start with a limited rollout. Give the chatbot to your top 20 customers and ask for feedback. Tune the responses based on what customers actually ask (which will be different from what you predicted). Expand gradually as accuracy and customer satisfaction improve.
The ROI Case
If your 5-person sales team spends 30% of their time on routine information requests, that is 1.5 FTE equivalent of time. At $80,000 fully loaded cost per sales rep, you are spending $120,000 per year on activities that a chatbot can handle. A well-implemented AI chatbot costs $12,000 to $36,000 per year depending on the platform and integration complexity. The ROI is 3x to 10x in direct labor savings, plus the revenue upside from your reps having 50% more selling time.
Your customers are already interacting with AI in their personal lives. They order from Amazon, bank on their phone, and get customer support from chatbots at dozens of companies. Bringing that same convenience to their steel purchasing experience is not futuristic. It is expected.