How a Pune sheet metal manufacturer 6×'d quote throughput and stopped losing deals to faster competitors.
Vajra Enclosures is a Pune-based contract manufacturer making custom sheet metal enclosures — electrical panels, telecom cabinets, HVAC housings — for OEMs across India and Southeast Asia. 220 employees, ₹78 Cr annual revenue, two plants in Chakan and Pirangut. 45–60 RFQs every business day, mostly arriving as emails with attached PDF drawings.
Sandeep Marathe, Head of Inside Sales, was watching ₹40–50 lakh of pipeline slip every month because his team couldn't quote fast enough.
A typical RFQ landed as a 3–15 page PDF with a 2D drawing, material spec, finish requirement, and quantity. The 6-person inside sales team had to read the drawing (45 min), check material inventory across two plants in SAP (20 min), wait 4–8 hours for the production planner to estimate labor and machine time, calculate margin against the customer's last quoted price, and get sign-off from Sandeep or the GM for anything above ₹3 lakh.
Average turnaround: 2.8 days. Win rate when they responded inside 24 hours: 41%. Win rate when they took 3+ days: 14%.
Schneider's procurement team sent an urgent 800-piece RFQ on a Monday morning. Vajra sent the quote Thursday evening. Schneider had already issued the PO to a Coimbatore competitor on Wednesday.
Vynara built a quote-generation system over 7 weeks. It reads incoming RFQ emails (Gmail + Outlook), extracts specs from PDF drawings using a vision model fine-tuned on Vajra's 4-year quote history (12,400 past quotes mapped to outcomes), cross-references SAP inventory, pulls labor and machine rates from the costing master, and produces a draft quote with full margin breakdown.
The estimator reviews the draft on a dashboard, adjusts if needed, and sends with one click. Complex drawing judgement stayed human. Data lookup, costing math, and quote formatting became automated.
The first week, I kept double-checking the AI's numbers. By week three, I was checking my own numbers against the AI's. It catches stuff my best estimator misses on a Friday evening.
Schneider came back in October. We quoted in 14 minutes. PO landed the same day. That one deal paid for the project twice over.
If you have 2+ years of historical quotes and structured costing data sitting in any ERP, this is a 6–8 week build. The estimators didn't get replaced. They got 6× faster and stopped working weekends.