Eliminating manual RFQ processing by introducing automation without adding headcount

The Challenge

One of the largest independent US-based industrial gas and welding supply distributors manually processed hundreds of RFQ emails every week. Requests arrived in different formats, including plain email text, spreadsheets, ERP-generated PDFs, and scanned images, all handled manually by customer service representatives. To turn a request into a quote, competitor part numbers had to be cross-referenced against the correct internal SKU before building the quote from scratch. There was no automated path connecting the inbox to a finished quote, so quality and speed varied with each rep’s experience and institutional memory. 

The Solution

Highspring partnered with the distributor to design and build a production-grade AI quoting pipeline that takes an RFQ email from inbox to a formatted draft quote without manual intervention.  

The system ingests RFQ emails across all four formats customers typically submit: email body text, XLSX purchase history, ERP requisition PDFs, and scanned images processed through optical character recognition (OCR). Once parts are identified, the pipeline runs a five-step progressive cross-reference lookup against the client’s ERP system, resolving competitor and vendor part numbers to the correct internal SKU. The institutional knowledge that once lived with a handful of experienced reps is now encoded directly into the platform. 

Three targeted AI capabilities power the pipeline working in concert. Email intent classification across nine categories ensures each message routes correctly immediately upon receipt, with part number and quantity extraction pulling clean, structured data from customer writing. When the ERP returns multiple candidates, LLM-driven disambiguation selects the right match without stalling the process. 

The system automatically produces a quote in the client’s standard template before drafting the reply, reviewing the draft, and then sending. The entire pipeline is production-deployed on Azure with enterprise-grade security, CI/CD, and full observability, giving the client’s technology leaders the governance, reliability, and visibility they expect from a business-critical system. 

Our Impact

Highspring’s AI quoting pipeline eliminated the need for manual cross-reference lookups and delivered a fundamental operational shift.  

By turning a headcount-bound manual process into a scalable, data-driven pipeline, the distributor freed its team to focus on customer relationships and higher-value work and generated key outcomes including: 

• Elimination of time-intensive manual cross-reference lookup work 

• Automatic categorization of different customer requests  

• Volume growth absorbed by the platform without adding headcount typically required to scale