Berlin-based mika has raised €6 million to expand AI-native accounting and tax software for small German companies, applying automation to one of the more rules-heavy parts of running a business.

The seed round was led by Smedvig Ventures, with Wecken & Cie. and several angel investors joining the financing. mika says more than 750 companies now use the platform, up from around 400 at the beginning of 2026.

The company targets GmbHs and UGs that must keep formal accounts and meet recurring tax obligations but may not have large internal finance teams. Its software covers bookkeeping, VAT returns, annual accounts, and tax filings, while trained accountants employed by mika review results and handle cases where software alone is insufficient.

The operating model illustrates how vertical AI is beginning to blur the line between software and professional services. Instead of adding a chatbot beside an accounting product, mika is attempting to own the workflow through which documents and payments become bookkeeping entries, financial records, and tax submissions.

Accounting gives vertical AI a demanding test

Finance contains large quantities of repetitive information processing, making it an obvious target for automation. It is also an unforgiving environment because errors can affect tax liabilities, statutory accounts, cash flow, and records that may be examined years later.

mika describes its core as an AI-native general ledger in which different models interpret receipts and payments within accounting rules developed by the company. Human accountants review the work where complexity requires intervention.

The platform also gives customers conversational access to their figures and can explain correspondence from the tax authorities, while individual tax advice remains outside the core product and is provided through licensed tax advisers in its network.

That division is important. Artificial intelligence can reduce the effort involved in interpreting documents and applying routine accounting treatment without removing regulated professional responsibility from every part of the process.

The company says revenue has been growing by more than 20% a month and passed €1 million in annual revenue in July. Those are company-reported metrics, but they indicate that the product is gaining traction around an existing administrative burden rather than relying solely on enthusiasm for generative AI.

Software moves into professional services

Accounting is part of a wider market in which software companies are absorbing activities once delivered primarily through service organisations. Legal work, insurance, recruitment, tax, and compliance are all seeing products built around documents and rules that models can increasingly interpret.

The economics depend on how much human work remains once the common cases have been automated. Even where software handles most documents without intervention, trained staff are still needed for exceptions, quality control, customer questions, regulatory requirements, and unusual transactions.

That can still produce substantial efficiency if professional time moves away from routine data entry and towards the minority of cases where judgement adds value.

It also creates a higher standard for auditability. A business cannot simply attribute an incorrect tax filing to an AI model. The system needs to retain the source documents, accounting treatment, adjustments, reviews, and other evidence behind a final return.

mika’s hybrid model recognises that constraint by keeping qualified people inside the operation. Its proposition is therefore less about eliminating accountants than changing the proportion of their time spent on routine processing.

The €6 million round will allow the company to develop the platform and expand further across Germany’s small-company market, where local accounting and tax requirements have historically limited the reach of international software products.

The harder test comes as the customer base broadens into businesses with more unusual transactions and less standardised financial activity. Vertical AI becomes commercially durable only when the system continues to produce dependable records after the straightforward cases have been automated.