An AI colleague for your firm: qualifying intake without partner hours
Nova conducts the first conversation: what is going on, which area of law, is there a deadline in play, and which documents already exist. That intake is structured and always the same, so a partner only sees files that are complete. What Nova does not do is judge whether someone has a case — that is legal judgement and it stays with the lawyer.
Published on · Last updated on
The problem in this sector
Partners only want conversations with qualified leads. Today every intake goes past a secretary first.
| Aspect | For this sector |
|---|---|
| Channels | |
| Own knowledge | Specializations, intake criteria, fee structure |
| Deployment | Nova Vault |
| Expected outcome | Every intake arrives with a lead score attached, so partners pick which conversations they take themselves. |
How far your sector has come
Legal practice falls under specialist business services at CBS, which uses AI relatively often — above average together with financial services, but below the 54 percent of information and communication. Note what that figure does and does not say: it measures whether a sector uses AI, not for what. In a firm that is more often about the administration around the work than about the legal work itself.
Where the limit sits
The line sits sharper here than in other sectors, for two reasons. An intake conversation almost always contains sensitive or confidential information, and it touches the professional secrecy lawyers are bound by. On top of that comes article 22 GDPR: turning someone down as a prospective client is a decision that directly affects them, and it may not be made solely by automated means. So Nova sorts and gathers; the decision whether to take a case remains a human one.
Concretely, these are the things Nova does not do on its own here:
- Judging whether someone has a case, how strong it is, or what it will cost.
- Turning down a prospective client — that is a decision that directly affects them.
- Sharing substantive documents or case information, not even with the person concerned.
That limit does not live in a setting someone can flip by accident; it is established up front. That is the difference between a system you know the behaviour of and one you find out about afterwards.
Confidentiality reaches further than the privacy rules
Alongside the privacy rules sits the duty of confidentiality in article 11a of the Dutch Advocates Act, and it reaches further. Everything you learn in your profession falls under it, even when it contains no personal data at all: a dispute between two companies is just as confidential.
That shifts the question. Not whether the privacy rules allow it, but whether an outside party sits inside your circle of confidentiality and who lays eyes on the correspondence along the way. Breaching it is also a criminal offence under article 272 of the Dutch Criminal Code — a regulator’s fine is not the only thing you are watching for.
In practice: the logistics can be automated — appointments, availability, which documents already exist. Anything touching the client’s legal position stays lawyer’s work. And whatever you automate belongs in your engagement letter, so the client knows before signing.
Written out further in AI at a law firm, down to what you record separately with the party you bring in.
The general rules on automated decision-making and transparency are set out in AI and the GDPR for SMEs and the AI Act for SMEs.
What we need from you
This differs per sector, so it sits here rather than in a generic checklist:
- The areas of law you do and do not take on, so the intake matches.
- Who runs the conflict check and when — that step stays human and belongs before substantive intake.
- The questions you always ask in a first conversation; that becomes the structure of the intake.
How a project runs — intake, analysis of your existing communication and a shadow week in which nothing is sent yet — is set out on the approach page.