Only 6% of Finance Leaders Want Full AI Autonomy

A EUR30 million round, a EUR1 billion valuation

Moss, the Berlin-based finance and spend-management platform, closed a EUR30 million (about $33 million) Series C round led by Portage, the fintech investment arm of Sagard, with participation from existing investor Cherry Ventures. The round values Moss at more than EUR1 billion, making it Germany's newest unicorn and one of the few European fintechs to cross that line in a funding environment that has stayed cautious through 2026.

The numbers behind the round are what make it more than a headline. Moss says it now serves more than 5,000 businesses across Germany, the UK, the Netherlands and Austria, generating annual recurring revenue above EUR70 million on 65 percent year-on-year growth. Its AI agents reportedly process more than 2 million transactions a month, handling the kind of invoice-coding and expense-categorization work that used to sit with a finance team's most junior hires.

Founded in 2019 by Ante Spittler, Anton Rummel, Ferdinand Meyer and Stephan Haslebacher, Moss plans to use the capital mainly to build new AI agents for finance workflows, with profitability targeted for 2027. That is a fairly ordinary funding story - fast growth, a credible investor, a large check. The more interesting part is one line most coverage buried well below the valuation.

The stat vendors do not lead with: 6 percent

Alongside the funding news, Moss cited a survey of 471 finance leaders that asked what they wanted most from AI in their finance function. Forty-eight percent ranked control as their top priority. Only 6 percent said they wanted full autonomy for AI in finance workflows.

That gap matters because it cuts directly against how most finance AI is sold. The standard pitch for two years running has been some version of "let the agent handle it" - autonomous invoice processing, autonomous reconciliation, autonomous approvals, with a human kept in the loop mostly for marketing copy. A survey this lopsided, 48 percent against 6 percent, is not a rounding error. It is finance leaders telling vendors, in a survey most of them will never see quoted back at them, that autonomy is not the feature they are buying.

Nobody writing about Moss led with this number. The valuation is the easier story: a fresh unicorn, a name to add to the European fintech list. But a funding announcement that quietly contains a market-research finding this specific, this measurable and this contrary to the industry's own sales pitch, is worth more to a buyer than the valuation is.

Moss built its pitch around control, not autonomy

Moss's own positioning tracks the survey almost exactly. Spittler has described the product as letting a finance team configure an agent for every finance job while retaining full control over each step and decision the agent takes, rather than switching on an autonomous system and trusting it. That is a deliberate contrast with competitors whose pitch centers on how much a business can hand off.

Whether that positioning reflects genuine product depth or careful messaging timed to a survey Moss commissioned is a fair question, and outsiders cannot fully answer it from a funding announcement. What can be said is that the company chose to frame a EUR1 billion round around configurability and oversight rather than around how much manual finance work its agents eliminate - a framing that would have made little sense if the market wanted autonomy.

That choice is itself informative, separate from whether Moss executes on it well. A company raising at a EUR1 billion valuation has strong incentive to say whatever it believes buyers want to hear. If Moss's read of the market is that control sells better than autonomy, that is a data point about where finance AI demand is actually heading, independent of Moss's own results.

What this means for anyone buying finance AI right now

Treat this announcement as a procurement signal rather than industry trivia. Any owner or finance leader currently evaluating a spend-management or finance AI vendor should use the 48-versus-6 split as a scoring framework: does the vendor show, not claim, approval-workflow granularity down to the transaction level; does every automated action leave an audit trail a human can review without exporting logs by hand; and how much friction does it take to override or reverse a decision the agent already made.

This is also a useful check for the UK reader specifically, where Moss already operates alongside domestic spend-management incumbents: ask any vendor, not just Moss, to show a live approval queue and an audit trail during the sales demo, not just describe one in a slide. A tool that cannot demonstrate override friction in the room is asking for trust the survey says finance leaders are not ready to give.

Vendors will keep pitching autonomy because it is the more exciting story to sell. Buyers evaluating finance AI in the second half of 2026 should score for control instead - not because autonomy is bad, but because the finance leaders who will actually sign the contract have already said, in a survey their own industry published, which one they want.

Read next: 95 Percent Smaller Models Change Where AI Runs  |  A Body Scanner Out-Raised Europe's AI Labs