Investors Bet 2.9 Million Euros on Plinto’s Invoice Agents

Copenhagen’s Plinto closed a €2.9 million pre-seed on October 20, 2026, to put AI agents on the invoices companies already book in their ERPs. People Ventures and Emblem led the round, with a group of business angels joining, on a bet that software which records every payment still leaves the actual checking to people.

The four Peakon alumni behind the company say those agents already flag errors in live books, then chase suppliers before the cash goes out. The raise is small. The claim is not: financial controls can run on every line, not on a month-end sample.

Investors Put 2.9 Million Euros on Invoice Agents

Plinto builds agents that sit on the systems a finance team already has, then test each transaction against contracts, budgets and internal rules. When something fails a check, the agents gather evidence, ping the right people and, if the customer allows it, write to a supplier for a credit note. A person still signs off on the action.

Nermina Suljevic is co-founder and CEO. Jakob Nielsen is co-founder and CDO. Sebastian Rehnby is a co-founder, and Andreas Garnæs is CTO. All four came out of Peakon, the Copenhagen employee-feedback company Workday agreed to buy in January 2021.

THE PRE-SEED

• €2.9 million: Pre-seed led by People Ventures and Emblem, with angels, closed October 20, 2026.
• Eight people: Four of them already sit in management seats, so the round has to buy builders and sellers.
• 2,700-plus banks: The stack also ties into Microsoft Business Central, Xero, Visma, Slack and Microsoft Teams.
• Four Peakon alumni: Suljevic, Nielsen, Rehnby and Garnæs, with paying customers and pilots already in the book.

People Ventures is a Copenhagen early-stage firm that started in 2016 as InQvation and backs Nordic software. Emblem is a Paris pre-seed firm that says it has raised $100 million to back Europe’s earliest companies, with a focus on France and the Nordics. For both, this is a wager that an action layer on top of the ERP can pay for itself in recovered cash rather than in prettier dashboards.

A Dashboard a CFO Would Not Trust

Nielsen studied economics and drifted into controlling after classmates talked about how hard it was to close the books. The first product was a business intelligence dashboard. A CFO looked at it and doubted the numbers. In late 2024, with new AI tools in reach, the team stopped trying to chart the books and started automating the checks that used to eat the close.

Plinto launched at the beginning of 2025. Nielsen spent much of that first year on site with finance teams and CFOs at a pilot customer, which is a dull origin story and a useful one. The product is being shaped by people who still close by hand, not by a lab demo of a chat window on a general ledger.

Companies have systems that record every single transaction, but checking whether those transactions actually match agreements and rules is still largely manual. We’re building AI agents that find deviations, investigate them and take action, so errors get fixed.

Jakob Nielsen, co-founder and CDO, Plinto

That sentence is the whole pitch. The ERP already has the posting. The gap is the matching, the follow-up and the credit note, work that still sits with controllers who sample the pile at month-end because they cannot read every line.

Agents Sit on the ERP and Write to Suppliers

Nielsen is careful about the category. He said Plinto is not an accounts payable suite. It is a layer in the background that talks to the people who already own each payment, so staff do not have to open another tool. Specialised agents split the work, watch one another, and more than one can be pointed at the same mess.

How far an agent may go is a customer setting. It can prompt a user, suggest a next step, or draft a note to a vendor. Nielsen’s line on that is simple: the human has control of what the AI does. The company site is blunter about catching mistakes before payment; the working rule, from the founders, is that posting and supplier messages stay under a person’s hand when the customer wants it that way.

WHAT THE AGENTS DO

• Line checks: Every sales order, purchase invoice and bank payment is tested against agreed prices, contracts, budgets and policy.
• Evidence pack: When a check fails, agents pull documents, delivery records and chat threads so a controller is not starting from a red flag and a blank screen.
• Internal chase: They ping buyers and budget owners in Slack or Microsoft Teams, including to confirm that goods actually arrived.
• Supplier chase: They can contact a vendor about a bad invoice, ask for a credit note, or prepare a correction for approval.
• Both rails: The same stack covers order-to-cash and procure-to-pay, and it does not replace the ERP it sits on.

Data is processed and stored in the EU, encrypted in transit and at rest, and the company says customer data is never used to train models. ISO 27001 certification is underway, not done. Integrations named so far are Microsoft Business Central, Xero and Visma, plus large Scandinavian banks and more than 2,700 other banks.

430 Discrepancies a Month in One Slice of the Books

Plinto has paying customers and pilots in logistics, real estate, facility services and food production, trades that run on thin margins and high document volume. In one slice of one customer’s business, the agents found around 430 discrepancies a month. The company did not say what those flags were worth, so that number is a count of issues, not a savings claim.

At another customer, the agents caught a duplicate invoice filed under a different supplier name in the same minute it arrived. That is the near-duplicate case exact-match ERP checks are built to miss: same obligation, different vendor record, second payment still unsent.

Suljevic put the commercial case in cash-flow terms. “Many of our customers work with low margins, so even small errors can quickly add up. The problem is that they’re often only discovered after the money has been paid. We want companies to catch and fix errors before they hit the bottom line,” she said.

The round is being retold as a Peakon reunion, which is true and also a dodge. The product only matters if 430 flags a month, and a duplicate stopped in a minute, show up again at the next customer, with a euro value attached.

Month-End Spot Checks Leave Cash on the Table

Large firms do not skip controls because they are careless. They skip them because the volume of postings is larger than a team that still works by sample. Controllers spot-check or wait for the close, and wrong prices, double invoices and mismatched deliveries clear the bank in the meantime.

Xelix, an accounts-payable analytics firm, went through 481 million invoices for a February 2026 leakage study and put preventable errors at 0.35% of annual spend, or $53 billion a year across the market it measured. That mix includes duplicate payments, invoice errors, missed credits and fraud. Exact-match rules miss transposed digits, extra suffixes on a vendor name, and the same bill arriving through two channels.

THE LEAK IN ACCOUNTS PAYABLE

A separate Xelix look at more than 1.2 million invoices across 187 ERP systems found duplicate payments still averaging 0.2% of supplier spend, and 0.22% of supplier spend even on SAP S/4HANA. Xelix’s own example: a business spending £8 billion on that stack should expect about £17.6 million in duplicate payments. Recovery audits, when they come in later, usually work the fattest suppliers and take a cut of what they claw back, which is a poor substitute for stopping the second payment.

That is the hole Plinto is selling into. If the leak is a fraction of a percent of spend, a low-margin logistics or food group feels it as profit, and a tool that only reports the damage after the close is too late.

A Munich Startup Is Already in the Same Lane

Suljevic said she has not met a direct rival in financial controlling, and she groups nearby products as AP tools, ERPs and general models such as Claude. The map is less empty than that sounds. In July 2026, Munich’s Finto raised a $3.4 million seed from Y Combinator, Gradient and Lightspeed for agents that check invoices, code them, match them to purchase orders and prepare ERP postings in SAP, Microsoft Dynamics and DATEV, handing the unclear cases to an accountant.

That is not the same product. It is close enough that a finance chief shopping for “agents on invoices” will see both in a week. Finto is built around European industrial ERPs and the posting itself. Plinto is built around deviation, evidence and the chase, including a note back to the supplier, on Business Central, Xero and Visma.

THREE NEARBY BETS

Repodo, started by Lunar’s founders, is a different bet again: audit work that still needs a qualified sign-off. Plinto is trying to sit before the audit, on every transaction, which is either a cleaner wedge or a harder enterprise sale. The person who still has to approve the agent’s draft is the part of this pitch that will matter to controllers who have already been burned by tools that posted first and explained later.

Eight Staff and a First Trip Outside Denmark

The money is meant for product work, more enterprise logos, and the first step outside Denmark. Suljevic wants product, engineering, sales and customer success lined up for a later round. With four of eight people already in management, hiring is the raise.

FROM PEAKON TO THIS ROUND

1. January 28, 2021: Workday agrees to buy Peakon for approximately $700 million in cash.
2. Late 2024: After a CFO rejects the dashboard, the team starts automating the checks that used to be manual.
3. Early 2025: Plinto launches in Copenhagen, and Nielsen spends the first year on site with a pilot finance team.
4. October 20, 2026: The company closes the €2.9 million pre-seed and publishes its pre-seed product plans.

Less than two years after launch, the test is not whether invoice leakage exists. Xelix already priced that. The test is whether an eight-person shop can turn today’s pilots into repeat contracts, with a euro value on those 430 flags, before larger AP vendors copy the chase-the-supplier layer and walk into the same mid-market accounts.

Disclaimer: This article is news reporting and analysis of a private company’s funding round and product claims. It is informational only and does not constitute investment advice, a solicitation to buy or sell securities, or a recommendation to adopt any software. Readers considering an investment in private technology companies, or a purchase of financial-controls software, should consult a qualified investment adviser or independent accountant who can review the primary documents. Figures, customer examples and product statuses are those given by the companies and research firms named above as of the dates on their materials, and they may change.