The debate over the real value of new LegalTech companies keeps coming back. One criticism is heard again and again: that they offer mere “wrappers”, superficial software layers built on top of existing large language models (LLMs). If the core functionality lies solely in the language model, the criticism may seem legitimate, especially with the rise of open source alternatives.
Yet this analysis deserves more nuance. There is a fundamental distinction between a simple technological feature and a viable commercial product, particularly when it is intended for professionals whose liability is at stake. This article examines the distinction between raw technology and the business application layer, and looks at why a solution’s packaging is, in fact, an essential component of its value.
What is an AI “wrapper” and why is the term used pejoratively?
In the field of artificial intelligence, a “wrapper” refers to an application that uses the programming interface (API) of a third-party language model, such as those developed by OpenAI or Anthropic, to provide a specific service.
The criticism leveled at these solutions rests on the idea that their added value is close to zero: if the language model is the tool’s only engine, the user could theoretically get the same result by querying the source model directly. In this view, the “wrapper” is seen as an empty shell, a presentation layer with no substance of its own, leaving the company vulnerable as soon as the language model provider evolves or cuts its prices.
Distinguishing raw technology from the commercial product: the packaging analogy
To understand why this criticism is incomplete, consider the analogy of an everyday consumer product, such as a chocolate bar. If the chocolate represents the language model (the raw substance), the packaging (the wrapper) plays a role that goes far beyond mere physical protection.
Packaging as a vehicle for trust and compliance
Quality packaging does more than contain the product; it fulfills critical functions that make the product marketable and safe:
- Brand identity: it guarantees a consistent experience and quality, regardless of where the product is purchased.
- Health and safety: it communicates nutritional information and allergen warnings, ensuring compliance with regulations.
- Accountability: it makes it possible to identify the company responsible if something goes wrong and to organize product recalls when necessary.
In the same way, in LegalTech, the “wrapper” should not be seen as a mere interface, but as the full set of mechanisms that turn a probabilistic technology into a reliable professional tool.
Packaging as an enabler of usability and mobility
Packaging also makes the product usable in a variety of conditions. It makes handling easier, protects the contents from outside elements (such as heat) and keeps things clean at the moment of consumption. For software, this translates into ergonomics, user interface design and the tool’s ability to fit into a working environment without creating technical clutter.
Why legal professionals demand commercial-grade “wrapping”
For lawyers and in-house legal departments, using AI cannot be reduced to a simple chat window. A solution’s value lies in its ability to meet the profession’s strict requirements.
Data security and confidentiality as the indispensable foundation
A commercial-grade product must address security and privacy issues that general-purpose language models do not handle natively. This includes:
- Regulatory compliance: GDPR compliance and certifications such as SOC 2.
- Protection of professional secrecy: the contractual guarantee that the data entered will not be used to train the underlying providers’ models.
- Control over infrastructure: the ability to offer secure, controlled environments, up to and including on-device processing.
Point of caution: Using a language model without an appropriate layer of protection and governance exposes the professional to risks of data leakage and breaches of ethical obligations.
Business process integration and domain expertise
A genuine LegalTech product must speak the language of lawyers and fit into their existing workflows. This involves:
- Software integration: the ability to communicate with case management tools or internal databases.
- Support and training: human and technical guidance to master the tool.
- Solving specific problems: the ability to handle complex tasks such as conflict-of-interest checks or the management of ethical walls.
Why the last mile is where the real value lies
The investors backing the major LegalTech players are not betting solely on the power of the language model, but on these companies’ ability to master the last mile.
The language model provides the processing capability, but it is the specialized application that turns this capability into a usable outcome for a lawyer. This crucial stage includes building specific workflows, managing legal accuracy and integrating source verification. As the underlying technology (the “chocolate”) becomes a commodity, accessible and inexpensive thanks to open source, value mechanically shifts toward the application layer (the packaging).
Vertical specialization: addressing the complexity of law
A common mistake is to treat law as a horizontal market, similar to other industries. In reality, the legal sector is a vertical made of verticals. It is composed of extremely deep specializations, whether in employment law, intellectual property law or public procurement law.
The solutions that succeed are those that offer vertical expertise:
- Domain depth: understanding the nuances specific to each area of law.
- Contextual accuracy: providing answers that respect the terminology and structures of each specialty.
- Adaptation to clients: meeting the distinct needs of law firms, notaries and corporate legal departments.
📌 Key takeaway: The value of a legal AI solution does not lie in the language model it uses, but in its ability to turn that technology into a specialized, secure tool that fits seamlessly into legal professionals’ workflows.
Conclusion
The “wrapper” criticism often confuses the technology with the product. While the language model is the indispensable engine, it does not by itself constitute a professional solution. For legal professionals, a solution’s importance lies in its application layer: its ability to guarantee security, uphold professional ethics, integrate into working processes and deliver sharp domain expertise. As access to language models becomes commoditized, it is precisely the quality of this “packaging” that will become the main factor of differentiation and trust in the LegalTech market.