Epona

Menu

Your DMS is full of junk. That’s why you need agents.

Inside Legal Platforms, by Bart Bogaerts

Legal platforms are about turning messy data into usable knowledge with agents

Most legal DMS environments are messy. That is not a criticism; it is simply the reality of how legal work evolves. Years of documents, emails, versions and matter files accumulate in structures that once made sense but rarely remain consistent in practice. For many firms, this is seen as something that needs to be fixed before AI can deliver real value. But the opposite is closer to the truth.

Within that apparent disorder sits precisely the information AI depends on to become useful. Not a perfectly curated dataset, but the full context of how work is done: communication, decisions, iterations, exceptions and precedent. That is what defines a legal practice. The issue has never been a lack of data. It has been the lack of a practical way to use that data at the moment it matters.

From answers
to execution

The first wave of AI in legal focused on answers: ask a question, upload a document, receive a summary or a draft. It created visible efficiency but remained fundamentally reactive. The user initiated the task, the system responded. The next phase shifts that dynamic. Legal work rarely consists of isolated actions. It is a sequence of steps: intake, analysis, coordination, review, decision-making and follow-up. Once technology begins to operate within that sequence rather than alongside it, the nature of its contribution changes. Instead of supporting individual tasks, it starts shaping how work progresses.

Context determines usefulness

The quality of output is no longer the defining factor. What matters more is whether the system understands the situation in which that output is used. A request to draft a document or review a contract is never standalone. It belongs to a matter, to a client, to a specific stage in a process, and to a set of prior decisions. Technology that operates without that awareness remains superficial, regardless of how polished the output appears.

What firms increasingly need is the ability to work with context: to understand what has happened, who is involved, what is relevant and what should happen next. That context exists at multiple levels, not only in matters and workflows, but also in how information itself is structured and organised within documents. That requires a structured environment in which that context exists and can be accessed.

Why representation
matters more than retrieval

Even when the right documents are retrieved, the outcome can still fall short. Not because the information is missing, but because the structure that gives that information meaning is lost along the way. Headings, speaker turns, table layouts and document hierarchy are not cosmetic details. They are signals that define how information is interpreted. When these are flattened into plain text, the AI receives something that is technically correct but practically unusable.

This is where many systems quietly fail. Retrieval metrics may suggest that everything is working, while the actual output remains vague, incomplete or difficult to verify. The gap only becomes visible when you look at the full chain, from the original question to the final answer and its citations. What matters is not only whether the right source is found, but whether the system can preserve enough structure to reason over it properly.

The revaluation
of the DMS

The DMS is valuable because it contains the complete record of how work is carried out. That record is rarely perfect as it includes inconsistencies, overlaps and gaps. But it also contains the detail that makes legal work meaningful: the progression of a matter, the evolution of documents, the reasoning behind decisions and the interaction between people. When that information becomes accessible within the flow of work, the role of the DMS changes. It moves from archive to foundation.

Knowledge management
without the illusion

Many knowledge management initiatives have struggled for a simple reason. They rely on consistent, ongoing input from professionals whose primary focus is delivering legal work, not maintaining knowledge systems. Documents need to be selected, cleaned, anonymised and structured. In theory, this is entirely logical. In practice, it is difficult to sustain.

What is changing now is the ability to derive value from existing data without relying exclusively on manual curation. By working with the information already present in matter environments and applying it in context, knowledge becomes more accessible without requiring perfect datasets.
This does not remove the need for governance or structure. It does, however, change where the effort is required and how knowledge becomes usable.

The role of the platform

For this to work, the underlying platform needs to provide more than storage. It needs to reflect how legal work is organised. That includes clear separation between matters, consistent access control, structured document relationships and reliable metadata. This is where SharePoint, and solutions built on top of it such as Epona365, become strategically important. Not as an additional layer, but as the environment in which legal work is structured in a way that supports both professionals and technology. The value comes from creating a coherent system in which information, context and workflow align.

What changes in practice

The impact of this shift is gradual but significant. Less time spent searching or reconstructing context, and more immediate access to relevant information. Matters that provide insight into their own progression rather than acting as static collections of files. Work that is less dependent on individual memory and more supported by the system. This allows legal expertise to be applied more effectively, by reducing the operational friction around it.

It is tempting to look for progress in new tools. The more important question is whether the existing environment can support intelligent work. The firms that move ahead will not necessarily be those adopting the most AI solutions. They will be the ones who understand how to use what they already have, by turning data into context and context into action. That process starts with a system many firms have had for years, but are only now beginning to fully understand: the DMS.