AI-inzichten & Trends

Private equity in Dutch law firms: opening the door to the AI-first law firm

Beau Jonkhout

The Dutch Bar is opening the door to outside capital in law firms. What does that mean for AI, and who stays in control in an AI-first law firm?

Het Vredespaleis in Den Haag met klokkentoren en leistenen daken, gezien over een tuin met siergrassen onder een blauwe lucht met wolken

Private equity in Dutch law firms: opening the door to the AI-first law firm

At the end of September 2026, the Netherlands Bar Association (NOvA) made a notable turn. Until now, only lawyers may own a Dutch law firm. At the symposium 'Modern practice structures in the legal profession', held at the Ministry of Justice and Security, the NOvA presented a vision in which, in principle, every firm structure becomes possible, including limited outside investment. According to the Bar's general council, private equity investors, accountancy firms, foundations and legal expenses insurers are not excluded in advance.

That is more than a governance question. The NOvA says so itself: outside investment can support investment in digitalisation and AI, as long as the risks to independence, control and confidentiality are managed. That brings a type of firm within reach in the Netherlands that already exists elsewhere: the AI-first law firm.

What the NOvA proposes

The core of the vision, as published by the NOvA on 29 September:

  • The 'staak'. All lawyers in a firm together form a separate organisational unit. This unit keeps decisive control over legal practice, the core values and professional decision-making. Outside investors may not gain decisive influence over the lawyers' work.
  • Prior review and a register. Firms that attract outside capital or work with other professions are reviewed in advance by the NOvA. Admitted firms are entered in a central register.
  • The firm itself becomes accountable. Besides the individual lawyer, the firm becomes subject to the professional rules, and can therefore be held to account by the disciplinary court on the core values: independence, partisanship, expertise, integrity and confidentiality.
  • Supervision of the organisation. Supervision also covers actual control and the role of investors. For that, the NOvA wants the independent legal profession supervisor (OTA) to be established as soon as possible.

This is not yet regulation. The NOvA is working out the conditions in phases, including a check against competition and EU law, followed by the register and the legal framework. The State Secretary intends to take a position this autumn on the research into alternative business structures in the legal profession.

That the Netherlands has a choice here has been settled since late 2024. In Halmer (C-295/23), the Court of Justice of the EU ruled on 19 December 2024 that member states may bar purely financial investors from law firms to protect lawyers' independence. They are not required to, and the NOvA is now choosing a controlled opening.

Why capital and AI belong together

The classic partnership model is poorly suited to large technology investments. Profits go to the partners every year, and an investment that only pays back after three or four years is partly paid for by partners out of their own pocket, for a return they may never see.

An AI-first firm needs exactly that kind of upfront investment. Think of knowledge bases holding the firm's own templates and precedents, workflows that connect intake, research and drafting, security that fits legal professional privilege, and quality assurance that shows a lawyer checked every result. Then there is the business model. A firm that bills by the hour and works faster with AI sees its revenue fall. Moving to fixed fees or subscriptions takes time and requires a buffer.

This plays out in a market made up mostly of small firms. Of the 5,494 Dutch law firms, 85.4 percent have four lawyers or fewer (NOvA, De Balie in Beeld 2026). For those firms, building their own AI is rarely feasible. Outside capital makes consolidation possible: franchise models, shared service organisations and firms built around technology from day one.

What an AI-first law firm is

An AI-first firm is something different from a firm that also uses an AI tool. Its processes are designed around AI: the software does the groundwork, and the lawyer assesses, decides and remains responsible. Where it is allowed, such firms already exist:

  • England and Wales. Outside ownership of law firms has been allowed there for more than ten years. In May 2025 the regulator, the SRA, authorised Garfield.Law, the first firm to build its legal services on AI. The firm helps small businesses recover small debts. Before approving it, the SRA looked at quality control, confidentiality, conflicts of interest and the risk of made-up answers. The system only takes a step when the client approves it, and named solicitors remain accountable.
  • United States. Arizona abolished its ban on non-lawyer ownership as of 2021. On 27 February 2025, KPMG Law US became the first Big Four organisation licensed there, with technology-driven services at its core.

The pattern is the same in both cases: the regulator allows the model, but only with human responsibility built in.

The 'staak' also decides about AI

This is where the Dutch proposal matters for anyone working with AI. An investor wants returns, and AI is the fastest route to efficiency. That is exactly why decisions about AI belong to professional decision-making, and therefore to the 'staak'. Firms considering an investor would do well to settle the following in advance:

  1. Who decides how AI is used? Agree that the lawyers' unit decides which tasks AI performs, which tools are used and how results are checked. An investor can provide budget, but cannot decide that a draft leaves the firm unchecked.
  2. Where does client data stay? Matter data should not end up with the investor or other companies in its portfolio, and should not be used to train models without an agreement.
  3. Who owns the knowledge bases and workflows? The knowledge bases and workflows a firm builds are a large part of its value. Agree what happens to them when the investor exits.
  4. Is quality assurance demonstrable? If the firm itself can be held to account by the disciplinary court, it must be visible that a lawyer checked the results and that sources can be traced.
  5. Does the business case rest on more than fewer hours? A plan based only on cutting hours puts the core values under pressure. A plan based on better and more affordable services fits the goal named by the NOvA and the WODC research: better access to justice.

And firms that are not looking for an investor?

For most firms, nothing changes in ownership for now. They will, however, face competitors with the capital to use AI at scale. The good news is that working AI-first does not require an in-house development team. Much of what an investor would fund is available as a service: legal AI with its own knowledge bases, source references and fixed review points.

At Prudai we build LEO on that principle. Each of LEO's 73 workflows has a fixed point at which the lawyer reviews and decides. The software does the groundwork and the lawyer stays in control. It is the same principle the NOvA now proposes for the firm as a whole.

What to expect

  • The NOvA is working out the conditions for collaboration, financing and governance, followed by the register.
  • Implementation is tied to the arrival of the independent legal profession supervisor.
  • The experiment with legal expenses insurers ends on 1 January 2028, which puts pressure on the pace.
  • The WODC research of April 2026 sets out a three-phase route to 2032, with outside capital in the second phase.

The direction is clear: the Dutch legal profession is making room for capital, and therefore for firms that use AI as their foundation. The question for every firm is not whether AI will change its work, but who stays in control when it does.

Sources

AIAgentic AIAI in organizations

Beau Jonkhout

Technical Director

Beau is co-founder and technical director of Prudai. He is the driving force behind the technical architecture of the Prudai platform. He leads the development of the multi-agent frameworks, manages the developers, and is responsible for the integration quality, security, and privacy by design of all solutions.