What one insurer’s migration reveals about the shift legal is about to make: from systems that record the work to a system that runs it.
There is a quiet moment in every large legal department that tells you more about the future of legal work than any conference keynote. It is the moment the old system gets switched off.
Earlier this year, the legal department of a large European life insurer reached that moment. For years it had run on a well-established desktop system for legal departments, good at what it was built for, which was filing matters and managing deadlines. When the vendor reorganised its portfolio around a new product line, a change of system became unavoidable. That is a common enough story in enterprise legal right now. What made it interesting was the question the department asked next.
They did not ask which archive to move into. They asked whether the move was a chance to work differently.
That question marks a shift the whole legal function is starting to make. Finance made it years ago, when spreadsheets gave way to systems that run the close. Sales made it with the CRM. In both cases the function stopped keeping a record of the work on the side and started running the work inside a system built for it. Legal has been the slow adopter, and the reason is not culture. It is that legal work is hard to systematise: every matter is a little different, the rules are exacting, and the cost of a quiet error is high.
The current wave of legal AI does not close that gap. Assistants that draft and summarise are genuinely useful, and every legal team should have them. But an assistant waits for a prompt and then hands the work back. It sits next to the matter; it is not the place the matter lives. It is also easy to replace with the next one. The shift that actually changes how a department operates is different in kind: the function getting a system it runs on, end to end, one that reads the incoming matter, runs the workflow, drafts the correspondence, keeps the record, and reports on the whole of it. An engine, not an assistant.
That does not make the assistants the enemy. The better model is to let them drive. Because the platform is built to be orchestrated, a lawyer can ask ChatGPT, Claude or Copilot to pull a deadline, summarise a matter or draft a response, and the engine underneath does the work and keeps the record. The assistant becomes the interface; the system of record and the intelligence live in the platform. The two worlds merge, with one of them doing the work.
This is not a thesis we arrived at from the outside. JUNE grew up where the pressure on legal is highest: thousands of parallel cases for Meta, for Europe’s leading airlines, for the legal and claims teams of major insurers, handled with full transparency and control. High-volume litigation is a hard school, because at that scale you cannot paper over a broken process with extra hours. The discipline that industrialises ten thousand cases turns out to be exactly what makes the single matter reliable. A function that can run at that scale can run everything below it.
For European legal leaders the first question is never the demo. It is data protection, and rightly so. Automation that a legal team can actually adopt, rather than pilot forever, has to be governed at the foundation: an EU-hosted, sovereign infrastructure, GDPR and DSA compliant, with the AI acting only inside the governed system, on controlled data, with a full audit trail behind every step. Trust is not a feature you add at the end. It is the thing that decides whether any of this leaves the sandbox.
Which brings the story back to that legal department. It did not just migrate; it moved its everyday work into one system. Incoming matters now run through workflows, the post is read and routed by AI rather than sorted by hand, correspondence is drafted in place, and the state of the department is visible in reporting instead of living in people’s heads. The hardest part of the move was not the documents. It was the department’s own receivables, interest-bearing claims with years of history, the kind of data where a single miscalculation surfaces in front of a court. Those came across without their history breaking, reconstructed and verified account by account, and they now sit inside the same file as everything else rather than in a spreadsheet beside it. The old system is switched off. The shadow spreadsheets are too.
None of this is really about one insurer, or one product. It is about a question that every head of legal will answer over the next few years, whether deliberately or by default: what does my function actually run on? For a long time the honest answer was “a filing system, plus the knowledge in a few people’s heads, plus a lot of manual work in between.” That answer is starting to look like the exception rather than the norm.
At the 9th World Legal Tech & AI Summit in Berlin this September, our CTO Christian Häp takes this further in his session, “Mission-Critical Legal: Why Departments Need an Operating System, Not Another Assistant”, from what high-volume litigation taught us about running legal end to end, to what it means to control one operating system from anywhere. If you leave asking what your own function runs on, it will have done its job.
JUNE is the operating system for legal work, an end-to-end, EU-hosted platform that runs high-volume legal processes on its own deeply integrated AI, from intake to resolution. Trusted by Meta, Europe’s leading airlines, and major insurers.
Learn More: https://www.june.tech/
By Katja Nikolaus, Co-Founder & CBDO, JUNE
