The legal AI market has produced some impressive companies. Harvey, valued at $8-11 billion, raised one of the largest AI rounds in history. Legora reached $1.8 billion. Noxtua secured exclusive rights to premium German legal commentaries. CASUS launched with Swiss market ambitions.
All of them are building the same thing: legal AI for legal work.
That is a problem. Not because legal AI is not valuable. It is. The problem is that professional work in regulated industries does not stay in one lane.
The Domain Boundary Problem
A corporate lawyer drafting a merger agreement will encounter tax provisions, employment law obligations, competition law thresholds, and potentially banking regulation. The legal work is inseparable from the tax work, the compliance work, and the regulatory work.
A compliance officer assessing a new product launch needs banking law, consumer protection regulation, data privacy requirements, AML obligations, and potentially insurance regulation. The compliance question is inherently multi-domain.
A tax advisor structuring a cross-cantonal reorganization needs corporate law, tax law across multiple jurisdictions, VAT regulation, and stamp duty provisions. The tax analysis cannot be completed without legal analysis.
Single-sector AI tools handle the first domain well. The moment the work crosses into a second domain, the professional is back to manual research. The AI tool that saved them two hours on the legal question cannot help with the tax question that is embedded in the same document.
This is not a minor inconvenience. It is a fundamental architectural limitation.
How We Got Here
The single-sector approach made sense as a starting point. Building a comprehensive legal AI system is hard enough without also trying to cover tax, banking, medical, and insurance regulation. Harvey started with elite US law firms. Legora started with European legal research. Noxtua started with German legal commentaries. Each chose a beachhead and built deep expertise.
The problem is that the architecture behind these platforms was designed for one domain. The data models, the embedding strategies, the retrieval pipelines, the user interfaces are all optimized for legal work. Adding a tax module is not a matter of loading tax data into the same system. Tax data has different structures, different hierarchies, different citation patterns, and different relationships to other domains.
Harvey, recognizing this limitation, has begun describing itself as a “professional services” AI platform. But its architecture, its team, and its data infrastructure are legal-first. Bolting on tax or compliance capabilities to a legal-first platform is like adding a kitchen to a car. You can do it, but the result is neither a good car nor a good kitchen.
The Network Effect Across Domains
The real power of multi-sector AI is not just convenience. It is intelligence that emerges from connecting domains.
Cross-domain citation graphs. A Swiss law references other Swiss laws. But it also triggers tax consequences, creates compliance obligations, and may interact with FINMA regulations. A multi-sector system maps these cross-domain relationships. When a user retrieves a corporate law provision, they also see the tax implications, the regulatory requirements, and the compliance considerations. This is not possible in a single-sector system because it simply does not have the other domains’ data.
Contextual risk detection. When AI understands both corporate law and employment law, it can flag that a proposed share transfer clause might trigger mandatory employee consultation requirements. When it understands both banking law and data privacy law, it can identify that a new digital onboarding process needs to satisfy both FINMA’s identification requirements and the FADP’s data minimization principle. Single-sector tools cannot see these intersections.
Compounding data value. Each new domain added to a multi-sector platform does not just add its own value. It multiplies the value of every existing domain. Legal becomes more valuable when connected to tax. Tax becomes more valuable when connected to compliance. Compliance becomes more valuable when connected to all of the above. This is a network effect that single-sector players cannot replicate.
The Economics of Multi-Sector
There is a hard business case here, not just a product argument.
For the platform builder:
A single-sector legal AI company builds its entire infrastructure, including vector database, embedding pipeline, retrieval system, hosting, compliance framework, multi-tenant architecture, and user interface, to serve one market. If it wants to add tax, it needs new data sources, new domain expertise, and significant re-engineering. The cost of the second sector is almost as high as the first.
A multi-sector platform builds the shared infrastructure once. Each new domain plugs into the existing stack: the same database, the same embedding pipeline, the same hosting, the same compliance framework. The marginal cost of adding a sector drops to roughly 30% of building the first one. By the fourth or fifth domain, the cost advantage is overwhelming.
For the buyer:
A Swiss bank that needs legal AI, tax AI, compliance AI, and regulatory monitoring today faces a choice: buy from four separate vendors (Harvey or equivalent for legal, a tax tool, a compliance tool, a regulatory monitoring service) and somehow integrate them, or buy from zero vendors because the integration burden is too high.
A multi-sector platform sells one subscription that covers all four needs. One login, one interface, one data governance framework, one vendor relationship. The total cost is lower, the integration burden is zero, and the cross-domain intelligence is included.
This is why enterprise software tends toward platforms rather than point solutions. The customer wants fewer vendors, not more. The institution wants one AI governance framework to manage, not four.
The European Opportunity
This dynamic is particularly acute in Europe, and especially in Switzerland.
European regulatory environments are inherently multi-domain. Swiss law interacts with EU directives through bilateral agreements. FINMA regulation implements international standards (Basel III, FATF) within a Swiss legal framework. Tax treaties between Switzerland and other countries create obligations that span corporate law, tax law, and international regulation simultaneously.
A US-focused legal AI company can serve US law firms that primarily work within one legal system. A European legal AI company must already handle multiple jurisdictions (Swiss federal, 26 cantonal, and EU). The step from multi-jurisdictional legal AI to multi-sector AI is smaller in Europe than anywhere else.
Yet the European market is the one where single-sector AI dominates. Harvey is US-centric (expanding to the UK). Legora is European but legal-only. Noxtua is German-speaking but legal-only. Nobody is building the multi-sector platform that the European market actually needs.
What Clients Are Starting to Demand
The market signal is already visible if you know where to look.
Law firms are asking their AI vendors about tax capabilities. Compliance teams are asking about legal research integration. Banks are looking for platforms that cover regulatory change management and legal analysis in one tool.
The common thread: professionals want AI tools that match how they actually work, which is across domains, not within artificial sector boundaries.
A recent survey by the Swiss Legal Tech Association found that 67% of Swiss legal professionals who use AI tools reported encountering situations where the tool could not help because the question crossed into a non-legal domain. The most common adjacent domains were tax (cited by 78% of respondents), compliance (61%), and employment regulation (44%).
These professionals are not asking for a better legal AI tool. They are asking for a tool that understands the full regulatory landscape they operate in.
The 12-18 Month Window
The multi-sector regulated AI space is empty. Nobody is occupying it. This will not last.
Harvey has the resources to attempt a multi-sector expansion. Thomson Reuters has the data assets across legal, tax, and compliance. Wolters Kluwer covers multiple regulatory domains through its publishing business. Any of these could decide to build a multi-sector AI platform.
But they have not done it yet. Harvey is architecturally constrained by its legal-first design. Thomson Reuters is constrained by its legacy publishing infrastructure. Wolters Kluwer is not an AI-native company.
The window for a new entrant to build a multi-sector AI platform from scratch, with an architecture designed for cross-domain intelligence from day one, is approximately 12-18 months. After that, the incumbents will have caught up or acquired their way in.
Companies that establish multi-sector AI platforms during this window will have a structural advantage: cross-domain data, cross-domain citation graphs, and cross-domain user workflows that later entrants will need years to replicate.
Mont Virtua is building exactly this. Enclava is a multi-sector AI platform for regulated Swiss industries, starting with law and FINMA compliance, expanding to tax, banking, insurance, and beyond. Every domain connects to every other domain through shared infrastructure and cross-domain citation graphs. To see what multi-sector regulated AI looks like, visit enclava.ch.