How Visma's AI accounting portfolio is reshaping bookkeeping for founders and compliance teams

This article is for informational purposes only and does not constitute financial, tax, or legal advice. Consult a qualified professional for guidance specific to your situation.

Editorial note: Reviewed for accuracy by the Startup Finance Guide editorial team. Our editors cross-reference all claims against platform documentation, regulatory publications, and vendor disclosures. Last reviewed: 2026-07-17.

Visma, a Norway-headquartered business software group serving more than one million small and medium-sized businesses across Europe, has disclosed that it now runs over 600 active AI initiatives inside its accounting and payroll portfolio, according to a July 2026 Sifted report based on interviews with the company's chief technology officer and two portfolio managing directors. The disclosure lands as European agentic AI startups raised €6 billion in 2025 and are on pace to exceed that figure in 2026, per Sifted's own market tracker data.

The story matters for cross-border founders because the accounting compliance burden does not shrink as a company scales across jurisdictions. A US-incorporated startup with Indian operations, for example, must satisfy Internal Revenue Service (IRS) reporting requirements, Foreign Exchange Management Act (FEMA) rules administered by the Reserve Bank of India (RBI), and potentially Goods and Services Tax (GST) Council obligations simultaneously. Any tool that genuinely reduces manual data entry and improves audit trails is worth scrutinising, provided founders understand what the platforms actually automate versus what they still require human sign-off on.

What changed

Visma's AI push is not a single product launch. It is a portfolio-level strategy in which the group acquires accounting software companies with high product engagement, then identifies which AI features can be scaled across the wider group. T. Alexander Lystad, Visma's chief technology officer, describes four categories the group targets: automation (data extraction, automatic booking, bank reconciliation), assurance (anomaly detection in payroll), advisory (surfacing patterns for human review), and agentic (end-to-end workflow automation).

Two portfolio companies illustrate the approach in practice. Chaintrust, a French accounting automation startup now part of Visma, uses AI to scan supplier invoices, extract line items, match them to ledger accounts, and post entries without manual retyping. Mikael Gandon, Chaintrust's managing director, says the trigger for building the product was watching junior accountants spend evenings retyping invoices into production software. The company measures success not by lab accuracy scores but by automation rate: the share of entries processed without a human touching them.

Dinero, a Danish accounting software company also inside the Visma group, takes a different angle. Its virtual CFO feature connects directly to a business bank account via open banking, completes routine bookkeeping, and lets business owners query their financial data in plain language. Martin Thorborg, Dinero's managing director, says the system predicts whether invoices will be paid on time, flags over-reliance on a single client or supplier, and benchmarks performance against comparable businesses.

On accuracy, Dinero says it separates its AI stack by risk level. Generative models handle advisory outputs like the virtual CFO. Core accounting workflows use machine learning and deterministic algorithms specifically to avoid hallucination. The company runs approximately 20,000 automated tests on every update before release.

Visma is not alone in this space. Intuit's QuickBooks and Xero, both widely used by cross-border startups, have rolled out AI-assisted categorisation and reconciliation features over the past two years. Paris-based Pennylane, which targets European SMBs and accounting firms, raised €75 million in 2024 and has positioned its AI layer as a direct competitor to legacy platforms. Founders evaluating Visma's portfolio tools should run a parallel assessment of these alternatives, particularly for jurisdictions where Visma's local-market coverage may be thinner.

What this means for founders

If your startup is spending more than a few hours per week on invoice entry, bank reconciliation, or month-end close preparation, the category of tools Visma's portfolio represents is worth a structured evaluation. Here is what to check before signing a contract with any platform in this space.

First, confirm jurisdiction coverage. Visma's portfolio companies are built close to specific local markets, which is a genuine advantage in Europe but may mean limited support for US Generally Accepted Accounting Principles (GAAP) reporting, IRS Schedule requirements, or Indian GST filing formats. Ask the vendor directly which tax jurisdictions their automated posting rules cover, and get the answer in writing.

Second, audit the automation rate claim. Gandon's point about lab accuracy versus real-world automation rate is the right frame. A platform claiming 95% accuracy in controlled conditions may still require manual review on 30% of entries in production because edge cases cluster around exactly the transaction types that matter most for compliance: intercompany transfers, foreign currency invoices, and split-category receipts.

Third, understand the human-in-the-loop design. Dinero's separation of generative AI for advisory and deterministic algorithms for core accounting is a reasonable architecture, but it is not universal across the market. Ask any vendor which outputs require mandatory human approval before posting to the general ledger and which are posted automatically. For a startup facing a tax audit, the answer to that question determines your liability exposure.

Fourth, check data residency. Cross-border startups handling financial data across the US, India, and Canada face data localisation requirements under rules including India's Digital Personal Data Protection Act (DPDPA) and various provincial privacy laws in Canada. Where your accounting data is stored and processed matters for compliance, and cloud-based AI accounting tools vary significantly on this point.

Fifth, run a cost comparison against alternatives. QuickBooks Advanced, Xero's partner edition, and Pennylane all offer AI-assisted features at price points that may undercut a Visma portfolio subscription depending on your transaction volume and headcount.

Limitations and open questions

The Sifted article that prompted this explainer is sponsored content, meaning Visma paid for placement. That does not make the technical claims false, but it does mean the article does not include independent verification of Dinero's 20,000-test claim, Chaintrust's automation rates in production, or the share of Visma's 600-plus AI initiatives that have moved from experimental to scaled. Sifted has not published a separate editorial investigation of these figures.

More broadly, no major financial regulator, including the IRS, the Securities and Exchange Commission (SEC), the Financial Conduct Authority (FCA) in the UK, or the RBI, has yet issued formal guidance on the use of AI-generated accounting entries as a basis for regulatory filings. That gap matters. If an AI tool posts an incorrect entry that flows into a tax return or a statutory audit, the liability sits with the company and its directors, not the software vendor. Standard software contracts disclaim this liability explicitly.

The agentic accounting category, where AI completes entire workflows without human initiation, is the least mature of Visma's four tiers. Lystad describes it as a target state rather than a current product. Founders should treat any vendor claim about fully autonomous close processes with caution until independent auditors have reviewed the outputs at scale.

Finally, the competitive picture is moving fast. Embat raised €30 million in May 2026 for treasury management automation. JetHR raised €25 million in June for HR and finance automation. Neither is a direct Visma competitor today, but the pace of funding in this category means the feature gap between platforms can close or open significantly within a single fiscal year.

This article is for informational purposes only and does not constitute financial, tax, or legal advice. Consult a qualified professional for guidance specific to your situation.