At a glance

​

• Customer: Montanstahl AG – Swiss producer of stainless and special steel profiles

• Use case: AI agent for RFQ feasibility

• Data used: 27+ months of RFQs, quotes, drawings, emails and specs

• Impact: >85% accurate feasibility; answers in seconds instead of hours, with thousands of senior sales hours saved and low seven-figure $ annual upside (est.).

The Business Challenge: The "Feasibility" Bottleneck

Before a quote could even be prepared, Montanstahl’s sales team had to get a clear internal “go / no-go” on each RFQ:
 

• Can we produce this geometry in the requested grade and tolerance?

• Which route or mill should we use?

• What’s a realistic lead time and what alternatives can we offer?
   

For complex special profiles, those answers weren’t in a system. They were buried in:

​

• 27+ months of RFQs, drawings and technical emails

• local spreadsheets and internal notes

• the heads of a few senior sales and product experts.
   

Getting to a decision meant long email threads, calls across time zones, and waiting until “the right person” had time to look.

Impact on the business
 

• RFQ feasibility checks often took hours instead of minutes, slowing down quotes.

• A handful of senior people became a bottleneck for day-to-day “can we make this?” questions.

• New sales hires needed months before they could handle more complex RFQs on their own.

• High-value RFQs risked being delayed or dropped simply because feasibility was unclear.
   

The C-suite objective was simple:

​

Automate as much of the feasibility step as possible, without asking sales to change tools or workflows. 

The Solution: Mavlon AI

Mavlon was selected for one primary purpose: to provide the sales team with instant, accurate answers to "Can we do this?"
 

This was a fundamental re-engineering of the pre-sales workflow:
 

1. From hours to seconds: Mavlon processed and understood 27+ months of technical sales emails, product specs, and past inquiries. This data was transformed into an intelligence layer capable of understanding complex technical RFQs and instantly determining product feasibility, viable alternatives, and availability.
    

2. Seamless Workflow Integration: Mavlon was embedded natively within the sales team's existing communication tools. This strategic, "no-friction" approach drove 100% adoption. Sales managers now receive immediate, accurate feasibility answers directly within their standard workflow, eliminating the need to switch applications.
    

3. A "Data-First" Security Model: Trust and data sovereignty were paramount.

   • 100% Data Ownership: Montanstahl retains full control over its proprietary knowledge base.

   • Total Privacy: Their data is never used to train public models, ensuring full GDPR compliance and protecting their competitive advantage.

Results & Business Impact

The Mavlon AI now sits in the same tools Montanstahl already uses for RFQs and gives an instant, traceable feasibility suggestion for most incoming requests. 
 

• >85% accuracy on feasibility suggestions
  The agent consistently meets the internal >85% accuracy benchmark for “can we make this / which route / rough lead time?” decisions on supported RFQs. Sales still has the final say, but the first suggestion is almost always usable or only needs a small tweak.
   

• Seconds instead of hours for routine checks
  What previously required emails, phone calls and digging through old RFQs is now answered in seconds for standard and semi-standard cases. Only edge cases and genuinely new geometries go back to senior experts.
   

• Several thousand senior-sales hours freed per year (est.)
  Based on Montanstahl’s RFQ volumes and internal time-savings estimates, the agent is on track to free up thousands of hours of senior sales and product time per year—time that can be reallocated to complex deals and strategic customers instead of routine feasibility checks. This translates into a low seven-figure annual upside when you combine capacity gains with higher win rates on qualified RFQs.
   

• From tribal knowledge to a searchable “mill memory”
  27+ months of RFQs, quotes, drawings and emails have been turned into a secure, searchable memory of past decisions. Feasibility answers are now consistent, auditable and available to the whole team, not just to the people who remember a similar job from five years ago. 
   

• Faster ramp-up for new sales staff
  New hires use the agent as their “on-demand expert”. Instead of waiting months to absorb all the product and routing knowledge, they can confidently handle more complex RFQs in a much shorter time.
   

Together, this moves feasibility from a manual bottleneck to a repeatable, AI-supported step in Montanstahl’s RFQ process—without replacing expert judgment where it really matters.

Next Steps:

Together with Montanstahl, Mavlon is now moving the RFQ feasibility from decision support to a fully agentic autonomous workflow:
 

• Extend data sources beyond emails and RFQs to include stock, capacity, routing, and pricing data.

• Auto-qualify standard RFQs by having the agent read incoming requests, run feasibility checks, and suggest go / no-go + alternatives by default.

• Minimise manual touch so sales managers only step in for edge cases and complex, strategic deals.
   

End goal: routine RFQs handled automatically, senior sales focused on high-value opportunities.