Mercura Taught AI to Read the Tender and Write the Quote

There is a room inside every industrial distributor that never makes the brochure. It is where a request comes in - an email, a phone call, sometimes a scanned tender running dozens of pages - and someone has to read it, figure out which products in a catalog of thousands actually match, check the margins, and reply before a competitor does. Do that hundreds of times a week. That room is where deals are quietly won and lost, and it is the room Mercura decided to rebuild.

Mercura, a Munich company in Y Combinator's Winter 2025 batch, makes software that turns unstructured sales requests into structured, spec-compliant quotes and orders. It reads the messy input, matches it to a company's product catalog, prioritizes the higher-margin options, and pushes the finished quote straight into the ERP system the business already runs on. The pitch on its homepage is blunt: stop losing deals to manual quote and order handling.

The problem hiding in plain sight

Industrial distribution and manufacturing are not short on demand. They are short on time. A single request for quote can involve a tender document written in the dense language of engineering specs, referencing standards and part numbers that have to be translated into whatever the distributor actually stocks. The work is real, it is skilled, and it is also the kind of task that scales badly with people. Hire more staff and you get more quotes, but slowly and at higher cost.

Mercura's founders had a front-row seat to this. CEO Lukas Bock grew up in a family with more than a century in the construction trades - the kind of household where quotes were discussed at the dinner table. Before Mercura he worked as a data scientist, including a stretch at Google. The company he built is downstream of that memory: the boring, grinding paperwork that keeps a plumbing or HVAC business alive.

His co-founders sharpen the picture. Sean Sdahl, the CTO, is an International Physics Olympiad silver medalist who worked on engineering at Bosch and Mercedes-Benz AMG before joining. Stefan Zheng, the chief product officer, came from Bain and research stints tied to MIT and Harvard. Hai Dang rounds out the founding group with a background in human-AI collaboration. It is a lot of horsepower to point at spreadsheets and tender documents, and that is precisely the point: the messy middle of industrial sales is a harder machine-learning problem than it looks, because the input is written by humans who assume another human will read it.

How the machine reads a request

Strip away the marketing and Mercura is doing document understanding on hard mode. The specifications engine parses tender texts and links them to matching products and services. Email requests get processed automatically and exported into CRM and ERP. There are AI voice agents that can field a customer call, gather requirements, and produce a quote in real time. The connective tissue is a set of enterprise integrations into systems like SAP and Oracle, wrapped in ISO/IEC 27001 and 27018 certification and GDPR compliance - the boxes a European procurement team needs ticked before it lets software near its order flow.

The claim the company leans on is speed: customer requests handled up to five times faster. In distribution that number matters more than it sounds, because the first spec-compliant quote through the door usually wins the business. Compressing a multi-day turnaround into minutes is not a convenience feature. It is the difference between quoting and not quoting at all.

What is easy to miss is the margin logic buried in step two. A distributor rarely stocks one product that fits a spec; it stocks several, at different price points and different profitability. A tired salesperson working through a backlog will grab whatever is nearest to hand. Mercura's engine can weigh the options and surface the higher-margin item that still meets the requirement, including private-label lines a distributor would rather move. Done once, that is a rounding error. Done across every quote, every day, it changes the shape of a business's income.

The number that sells it

Mercura's most persuasive stat is not about the AI. It is about coverage. One distributor, the company says, went from answering roughly 70 percent of incoming requests to 100 percent - while improving customer satisfaction - without adding headcount. The requests that used to fall through the cracks, the ones that arrived at a busy hour and never got a reply, now get quoted. For a business that lives on volume, closing that gap is found money.

Who is actually buying

The customer list reads like a tour of European industry: named references include Bauder, Reisser AG, Siteco, and the BME Group, which spans BAUKING and SHK Deutschland. These are not startups experimenting with AI for a press release. They are distributors and manufacturers with real catalogs and real quotas, and the fact that 60-plus of them are on board is the strongest signal Mercura has. Enterprise buyers in this world are conservative by habit; they do not adopt software that fails in front of a customer.

How it is different

The obvious comparison is to legacy CPQ - configure, price, quote - tools. But those systems assume the input is already structured. You feed them clean data and they assemble a price. Mercura's premise is that the input is never clean. The request is a paragraph of email, a photographed spec sheet, a voice on the phone. The hard part, and the part Mercura is betting its business on, is the messy front end: understanding what the customer actually wants before anyone has typed it into a form.

There is also a quieter difference in how the product enters a company. Rip-and-replace software asks a distributor to change how it works, which is why so much industrial software never gets adopted. Mercura slots in behind the existing ERP and CRM rather than in front of them. The salesperson keeps their tools; the AI handles the reading and drafting and hands back a quote for a human to send. That posture - assistant, not replacement - is part of why conservative buyers say yes. Nobody is being asked to trust a black box with the final word to a customer.

The business, and the bet

Mercura sells B2B SaaS to distributors and manufacturers, priced around automating the quote-and-order workflow and integrating with the systems they already run. The unusual detail is the balance sheet: the company says it was profitable before raising any outside money, and is targeting double-digit-million turnover in 2026. When the seed round came - an oversubscribed $2.1 million-plus in December 2025, led by TQ Ventures and SignalFire - it arrived on top of a business that already worked.

The angel list tells its own story. Backers include Bastian Nominacher, co-founder of Celonis; Tao Tao of GetYourGuide; a senior figure at SAP; and entrepreneur Susanne Porsche. These are people who understand exactly how much slow, manual process is buried inside large industrial companies - and who apparently think Mercura is aimed at the right seam. Celonis, worth remembering, built a multi-billion-dollar business on process mining: the art of finding the wasteful, repetitive steps inside enterprises. Its co-founder writing a check into a company that automates one of those steps is not a coincidence.

The stated plan for the new money is unglamorous and probably correct: expand the engineering team, sharpen the product-matching engine, and push into new markets. There is no talk of a moonshot pivot. For a company that reached profitability by solving one specific pain very well, staying pointed at that pain is the harder discipline - and the more valuable one.

Where it fits

Zoom out and Mercura sits in a broad wave of software trying to modernize the parts of the economy that software mostly skipped. The flashy AI companies chase lawyers and programmers. Mercura went the other direction, toward the trades and the warehouses, where the tools are old and the problems are enormous. It is a less crowded street, and if the coverage numbers hold, a lucrative one. The team - which includes physics-olympiad medalists and former engineers from Bosch, Mercedes-Benz, Google and Celonis - is technically dense for a company most people will never hear of. That is usually a good sign in B2B: the harder the demo is to make look impressive, the deeper the moat when it works.

The risks are the ordinary ones for a company at this stage. Every distributor's catalog is a little different, every ERP install has its quirks, and the value of the product depends on matching being reliable enough that a salesperson stops double-checking it. Get that wrong and adoption stalls; get it right and each new customer makes the engine a little smarter. The certifications and EU hosting suggest Mercura understands that trust, not cleverness, is the currency in this market. Enterprise buyers forgive a slower roadmap. They do not forgive a quote that goes out wrong to their biggest account.

For now, the story is simple enough to fit on a business card. Requests come in messy. Mercura makes them come out as quotes. Sixty-plus companies that keep the world running are paying for exactly that, and a batch of investors who have seen this movie before are betting the number keeps climbing.