Accel Backs Atira to Automate Industrial Sales Engineering

Munich software company Atira closed a $15 million seed round led by Accel on September 3, 2026, taking total funding to $17.5 million. The cash is for AI that reads industrial requests for quotation, drafts technical bids, and sits on the CRM, ERP, and configure-price-quote tools factories already run.

Accel once led the Series A for Celonis, Munich’s process-mining giant. Celonis co-founder and co-CEO Bastian Nominacher is now writing a personal check into Atira, a company that is not yet two years old and already quoting live work at ABB’s charging unit, Chiron Group, and railway supplier Robel.

Accel Is Placing Another Munich Process Bet

The seed was led by Accel, with UVC Partners, Fortino Ventures, and Booom returning from a previously unannounced $2.5 million pre-seed. Industry names joined: Whirlpool chairman and CEO Marc Bitzer, former Chiron Group CEO Markus Flik, investor Ann-Kristin Achleitner, and Nominacher. Atira declined to disclose a valuation.

Accel partner Harry Nelis, who is based in London and has invested in software since 2004, said developing quotes for complex manufacturing “involves huge amounts of unstructured information, technical judgement and knowledge spread across different people and systems,” which ordinary software has failed to automate. “AI changes that equation because it can understand that context and orchestrate work across the entire process,” Nelis said.

That is the same firm that, in its own words, built Accel’s partnership with Celonis by leading the Series A and following on when process mining was still an academic niche. Celonis later raised a Series B at a $1 billion valuation. Nominacher sitting in Atira’s round is the tell that the people who sold factories on process data think quoting is the next pile of hidden labour.

Florian Diegruber, Atira’s CEO and co-founder, spent years as a commercial lead at Palantir, much of it inside one of Germany’s large carmakers. August DuMont Schütte, a former machine-learning engineer at Google, is chief technology officer. They started working together in mid-2024 on a broad idea about rebuilding industry in Europe and the United States, then narrowed to quoting. Diegruber said the category “sounded very unsexy back then, which is always a good place to start a venture.” They incorporated in Munich in November 2024.

The Front Door Still Takes Months

When a maker of fire trucks, rail gear, or EV chargers is asked to bid, the inbound file can run to thousands of pages. Sales engineers, specialists, lawyers, and commercial staff still pass that packet around for weeks, sometimes months, before a price leaves the building. Atira puts industrial sales engineering at more than $128 billion in yearly labour worldwide, a company figure, not a third-party census.

Diegruber’s case is that plants spent a decade automating the line and left the inbox alone. “Sales engineering is one of the largest workflows in manufacturing that software has never truly automated,” he said.

When you need six months for your sales cycle to go from request received to final proposal out, it’s great that you’ve automated a week or a month of your factory floor. But the big chunk is on the front door.

Florian Diegruber, CEO and co-founder of Atira

Aleran’s 2025 Built to Sell survey of manufacturers found that 86% of respondents said slow, manual quoting had cost them deals, and 70% said quotes still took more than a day. Those are vendor-sponsored numbers, and they match what machine builders already admit in private: the bid is often late, inconsistent, or skipped.

THE ASK-TO-BID CALENDAR

1. Mid-2024: Diegruber and DuMont Schütte begin working together on industrial software.
2. November 2024: Atira is incorporated in Munich.
3. November 2025: The product launches commercially.
4. September 3, 2026: Accel leads the $15 million seed, and the $2.5 million pre-seed is disclosed.

Ten months after launch, Diegruber said the company has about 15 customers, all in full production rather than pilots, and that five of those accounts have more than 100 users each. The commercial work between CRM and ERP is the gap the company says it was built to occupy.

What Atira’s Agents Do With an RFQ

The product is a set of agents that parallelize quotation work rather than a new system of record. Incoming RFPs, tenders, user requirement specifications, and mail are broken into structured requirements and pre-commented against the customer’s own product data and old remarks. A red-flag agent marks what the plant can fulfil and where it should push back. A bid/no-bid agent is meant to say, in the first pass, whether the team should spend the hours at all.

ATIRA’S PRODUCTION AGENTS

• Requirement extraction: Turns inbound packets into structured, pre-commented requirements.
• Red-flag detection: Marks clauses the plant cannot meet and where to push back.
• Bid / no-bid: Flags whether the team should invest hours or walk away.
• Norms and standards: Builds a report from industry, region, or a private norms library.
• CPQ support: Drives the customer’s configurator, price lists, or Atira’s own rules.
• Technical documentation: Drafts traceability matrices, compliance checklists, and bid packs.
• System writeback: Pushes results into existing CRM, ERP, and CPQ tools without a retype.

Tacit knowledge is the part no document holds. Diegruber said the agents ping human experts over Microsoft Teams when a file cannot answer, and that reviews show sources and require a person to confirm each one. “You can’t just deploy an agent and say the agent now does the work that you’ve previously done manually,” he said. He also said a generic chatbot does not scale across a sales force: “I always like to joke, if you give Claude to 20 sellers, you still get 20 different answers.”

The commercial claim underneath the agents is capacity. The same engineers clear more RFQs in parallel, and a bid/no-bid cut stops the team from burning a week on a tender it will never win. That is a volume story, not only a labour-saving one.

Chiron Processes Inbounds 80% Faster

Named logos on Atira’s site already span rail, machine tools, fire trucks, belt service, packaging machines, auto components, and energy tenders. Diegruber named ABB E-mobility, the charging business majority-owned by ABB, as a production customer. Chiron Group, the German machine-tool firm, has more than 70 people on the software and says inbound quote requests run 80% faster than before. Robel, which builds railway construction equipment, says the software saves 95 hours of sales and engineering time per quotation request.

Irfan Coskuner, team lead for tender projects, said his team evaluated several AI tender and requirement tools and that Atira “was implemented and working productively within a week.” Diegruber said the usual deployment is a day or two, with a one-month out, and that no customer has used the exit.

RESULTS ATIRA CITES FROM LIVE PLANTS

Rema Tip Top, Theegarten-Pactec, Fette Compacting, and Osbra are also listed as live. The 80% figure belongs to Chiron’s inbound handling, not to every account, and the 95 hours belong to Robel. Those are customer-reported operating numbers published by Atira, not a controlled study.

Nominacher and Bitzer Sat in the Round

The cap table is the wager in human form. Accel supplies the global distribution playbook it used when Celonis crossed the Atlantic. Nominacher supplies a founder who has already sold German process software into Siemens-scale accounts. Bitzer runs an appliance maker that lives on configured products and dealer quotes. Flik used to run Chiron, which is now a reference customer, so an operator who knew that quoting mess from the inside is funding the tool that claims to shrink it.

WHO PUT MONEY IN

• Seed lead: Accel, on September 3, 2026, for $15 million.
• Returning funds: UVC Partners, Fortino Ventures, and Booom, from the $2.5 million pre-seed.
• Industrial angels: Marc Bitzer, Markus Flik, Ann-Kristin Achleitner, and Bastian Nominacher.

Diegruber said most of the new money goes into engineering and the first commercial hires who are not founders. The company also said it will grow go-to-market staff, speed up product work, and push abroad, then extend the same agents into pricing intelligence, aftersales, and supplier coordination. Adjacent workflow is the expansion path; quoting is the wedge.

Why CPQ Never Swallowed the Paper Packet

Configure-price-quote software is not new. Tacton, founded in 1987, still sells constraint engines to machine builders, and it now argues that RFQ speed as a win-rate variable matters as much as the list price. SAP, Salesforce, Epicor, and a string of specialists already promise faster quotes for configured goods. Berlin-and-Munich rival Bryo pitches a similar AI layer for technical sales and says it works with the CPQ a plant already owns.

Atira is not trying to rip those systems out. It reads the unstructured front of the job (the 2,500-page energy tender, the red-flag clause, the bid/no-bid call) and writes back into CRM, ERP, and CPQ. Diegruber said consultancies often sell slow, expensive builds, and that other tools miss the unwritten rules inside a plant. Large software vendors and the major model labs are pitching the same industrial buyers. Other startups are in the hunt, including Roadrunner.

The open objection is not whether a model can summarise a PDF. It is whether a bid still dies because no named owner files before the window closes, and whether a second dashboard beside Salesforce creates another handoff instead of removing one. Accuracy is the other live risk. Atira’s answer is source-linked review and a human confirmation on every agent answer. That is slower than a fully automatic send, and it is the price of putting a draft in front of a customer who will hold the plant to the spec.

The First U.S. Customer Is Still in Pilot

HEADCOUNT AND REACH

• Staff at the start of 2026: Four people.
• Staff at the seed: About 15, matching the 16 listed on PitchBook and LinkedIn.
• Hires due: Nine more people scheduled to join.
• United States: One customer in a pilot, not yet in production.

Diegruber said European industrial buyers are more willing to take a startup’s software than the old stereotype allows, helped by a deploy-in-days model. The first American account is still a test. The German references are not. Robel is already counting 95 hours back per RFQ, and Chiron is running more than 70 people through inbound packets that used to sit in inboxes.

Accel has seen this movie in Munich before, when process mining left the university and became a budget line. Quoting is messier, because the input is a PDF mountain and the output is a promise the factory must keep. The seed pays for the engineers who have to make that promise hold at 100-user rollouts, not 15-person pilots.

Frequently Asked Questions

What Is a Request for Quotation in Manufacturing?

A request for quotation, or RFQ, is the packet a buyer sends when it wants a priced, technically valid offer for a machine, a plant, or a built-to-order system. In industrial sales it often arrives as a mix of PDFs, drawings, user requirement specifications, emails, and lists of norms, and public tenders can run for years with several departments scoring the reply. Private RFQs are usually faster but still pull sales, engineering, legal, and commercial staff onto the same file before anyone names a price.

What Does CPQ Software Do That Atira Does Not Replace?

Configure-price-quote software stores product rules, options, and prices so a salesperson cannot promise a combination the factory cannot build. It is strongest when the catalogue is already structured. Atira is aimed at the work before that catalogue is even in play: reading the inbound packet, marking impossible clauses, deciding whether to bid, and drafting the technical paper that later lands in CPQ, CRM, or ERP. Plants that already own Tacton, SAP CPQ, or a homegrown configurator are the intended users, not rip-and-replace accounts.

Does Atira Train Its Models on Customer Data?

Atira says customer data is never used to train AI models, that it is ISO 27001 certified, and that it is set up to meet GDPR rules. The company also advertises enterprise deployment options for plants that want the system inside their own security perimeter. Those claims sit on the vendor’s security page and have not been audited in this article.

Where Is Atira Headquartered?

The company is based at Sämannstraße 14, 82166 Gräfelfing, Germany, and is registered in Munich under HRB 298272. PitchBook lists 16 employees against Diegruber’s figure of about 15 at the seed, with nine further hires due. The product site is atira.ai.