The question was always the same: can we actually make this?
How the Montanstahl Group turned years of RFQs, drawings, technical emails, and production data from three sites into an instant RFQ screening engine, embedded directly in the sales team's existing workflow.
The problem every subcontractor recognizes
Before a quote could even begin, someone had to answer: can we produce this geometry in the requested alloy and tolerance? Which site and route should we use? What is a realistic lead time? What alternatives can we offer?
For complex profiles involving tight tolerances, aerospace-grade titanium, or unusual geometries, those answers were buried in:
- Years of past RFQs, drawings, and technical emails scattered across three production sites
- Scattered spreadsheets and internal notes in Stabio, Schwerte, and Talamona
- The heads of a few senior engineers who remembered similar jobs from years ago
The multi-site structure made it worse.
- Montanstahl AG in Switzerland handles cold-rolled and laser-welded profiles
- Montanstahl GmbH in Germany does hot extrusion in nickel and titanium
- Siderval in Italy produces hot-extruded seamless profiles in carbon steel, stainless, and superalloys
A single feasibility question often required cross-site phone calls that added days to what should have been a minutes-long decision.
Feasibility checks took hours instead of minutes while competitors responded faster. A handful of senior engineers became the bottleneck for every “can we make this?” question. New sales hires needed months before they could handle complex RFQs independently. High-value RFQs risked being delayed or dropped simply because feasibility was unclear.
The group's objective was simple: automate as much of the feasibility step as possible without asking sales to change their tools or workflow.
Instant screening of any RFQ
Mavlon's AI reads incoming RFQs and immediately determines: can the group produce this geometry, in this alloy, at this tolerance? Which production site is best suited? What is a realistic lead time? What alternatives exist if the exact request is not feasible? The system drew on the group's full production history and technical knowledge across all three sites, so a sales engineer in Switzerland got answers informed by Siderval's Italian extrusion capabilities without making a single phone call.
Zero-friction integration
The system was embedded directly into the team's existing communication tools. No new software to learn. No separate application to open. When an RFQ arrived, the feasibility answer appeared in the same workflow the estimator was already using. This drove 100% adoption from day one.
Full data sovereignty
The group retains full ownership and control of their proprietary data. Nothing is used to train external models. Full GDPR compliance.
85%+ screening accuracy
For 85% of incoming RFQs, the AI suggestion was directly usable or needed only minor adjustment. The remaining 15% were genuinely novel requests that correctly routed to senior engineers for human judgment.
Seconds instead of hours
Standard feasibility checks that previously required emails, phone calls, and cross-site coordination were answered instantly, including which production site was best suited for the job.
Thousands of senior hours freed per year
Repetitive “can we make this?” questions stopped reaching senior engineers. Estimated annual business impact from capacity gains and higher win rates: low seven-figure range.
New hires productive in weeks, not months
New sales staff queried the system as their on-demand expert instead of spending months absorbing tribal knowledge through questions and osmosis.
Consistent answers across all three sites
The same question asked by different people on different days produced the same answer. A sales engineer in Switzerland could instantly access Siderval's Italian production history without a single phone call.
“Experienced people answering feasibility one RFQ at a time, while the queue grows. Different questions. Same bottleneck.”
After “can we make this?” comes the harder question: what do we charge?
Montanstahl's question was whether a request could be made at all. For most custom manufacturers, the question after it takes far longer: what should the quote say? The answer sits in the same kind of places, a long bid package from the customer and the heads of the few senior people who know how the company prices.
That is the problem Mavlon works on now. Our AI Quoting Engine reads the customer's whole bid package, meaning the specification, the drawings, the bid form and the addenda, and turns it into a priced draft quote in the manufacturer's own pricing, learned from its past bid packages, the quotes it sent and its costing workbooks. Every line keeps the page it came from, a decision the documents leave open arrives priced both ways, and anything nobody can answer becomes a question for the customer. The estimator reviews the draft and sends the quote.
It is built for products that are designed for each order and arrive as a bid package. There are pages for switchgear builders, dock and gangway manufacturers and water and wastewater equipment makers, and worked examples for paralleling switchgear, e-houses, battery storage, lift stations, heat exchangers and pressure vessels.
See a priced draft quote for the work you build.
In a demo we show the AI Quoting Engine turning a real bid package into a draft quote, line by line, and how it would learn your own pricing from your past jobs.
Book a Demo