Managing Partner of the Aronov & Partners Bar Association on why a law firm needs a single, secure platform where legislation, case law, and documents are all in one place and accessible to the entire team.
Take a look at how work on a major case is organized in a law firm today. Contracts are stored in folders, correspondence is scattered across email and messengers, scanned documents are kept on a drive, and key facts are dispersed across dozens of files. To speed up workflows, lawyers unfortunately end up uploading some documents to AI chats and receiving answers that do not indicate which legal provision they are based on, which case materials support them, or why they can be trusted and relied upon, including in court proceedings. At the same time, case materials that should never leave the firm’s secure environment are sent to third-party servers abroad.
Working in compliance with current legislation without using AI is becoming increasingly difficult today.
At our bar association, we came to a simple conclusion. The problem is not that we need yet another smart tool — there are already too many of them, they are fragmented, and the case as a whole is not assembled anywhere. What we needed was not another neural network, but a single platform where the entire case — materials, facts, legal provisions, case law, and documents — could be kept in one place, securely and accessible to everyone authorized to work on it.
Why “just using an AI chat” does not work here
There are several reasons, and all of them are particularly important in litigation practice.
The answer cannot be verified. A lawyer cannot go into a hearing relying on a statement simply because “the neural network said so.” A source is essential: a specific document, a specific page, a specific quotation. If a conclusion cannot be checked against the original source, it is useless, no matter how accurate it may be.
Data cannot be transferred outside the local computer’s secure environment. Case materials contain personal data, trade secrets, and often information that must never leave the secure environment of the firm or the client. Sending such information to an external service, especially one that processes data outside Russia, is prohibited both by law and by the internal policies of most clients.
There is no full case context. A general-purpose model responds based on the text you paste into the window. It does not know that Document No. 47 supersedes Document No. 12 — because Document No. 12 was not included in the prompt. In a real case, however, what matters is the entire body of materials and how they relate to one another.
Legal errors come at a higher cost. An incorrectly paraphrased legal provision or a reference to a non-existent case is not merely an unfortunate inaccuracy; it creates the risk of adopting an incorrect legal position. The legal profession has already seen enough examples of this.
What we did: brought the entire case together in one place
That is why we decided not to wait for a ready-made solution to appear on the market and instead joined forces with the Agentic Lab team to develop one — and that is how Agata was created. The system was configured and tested using real cases, including arbitration disputes involving construction contracts and commercial matters. For the sake of transparency, our organization is not an independent observer here; we are both a co-developer and a user of the product. What has this delivered in practice:
A unified case workspace. All formats — PDF, Word, scans, photographs, and archives — are uploaded to one place. From there, the collection stops being just a mass of files: each document is associated with parties, dates, and amounts, while facts are linked across documents throughout the entire case. Any extracted attribute can be used as a filter. There is no need to keep the case open across ten different tabs — or to keep everything in your head.
Answers with source references. Questions can be asked in plain language, and every fact in the answer links to a specific document and page containing the relevant quotation. If something is not supported by the case materials, the system does not present it as fact. Unlike an “opaque chatbot,” here you can always see exactly where a conclusion came from.
Legislation and case law alongside the case itself. A legal provision can be checked in the version applicable on a specific date, together with its amendment history; you can also see how courts have decided similar disputes (beta). There is no need to switch to a separate legal research system and lose the context of the case.
Documents built on the actual case facts. Draft procedural documents — statements of claim, responses, objections, and motions — are generated not from an abstract template, but on top of the already analyzed case materials, using the parties, amounts, and dates from your documents and providing references to the relevant sources. The lawyer edits a draft instead of starting from scratch.
The team sees what it needs; the practice head sees everything
And this is precisely why a platform like this is needed by the firm as a whole, rather than by just one lawyer. A case in Agata is a shared workspace with roles and permissions configured for each individual case: owner, editor, view-only access, or access limited to a user’s own documents. Everyone working on the case sees the same materials and collaborates on them directly instead of emailing different versions back and forth. The practice head has a complete overview — the entire case portfolio, case stages, and team workload — while the activity log shows who opened or edited what. Case stages and case types can be configured to suit a specific practice area. Some collaboration features are currently in beta and are being actively developed.
Infrastructure and data
Perhaps the most important point for us as a firm is that the data does not go anywhere. Agata runs on servers located in Russia, within a secure environment, using proprietary models and proprietary OCR technology; it can be deployed either in the cloud or on-premise directly within the client’s infrastructure and is included in the Russian Software Registry. Case materials remain under the firm’s control — this was a prerequisite without which we would not have joined the project.
Security is one of the most important factors for us, which is why the system fully complies with the requirements of applicable legislation governing information protected as confidential by federal law, including personal data, as well as attorney-client confidentiality. None of the documents contained in the case materials ever leave the platform’s secure environment, where they are protected against unauthorized access and data leaks.
On quality
I will discuss quality without citing neat percentage figures, which in this field often serve marketing purposes more than they serve lawyers. Quality is always context-specific. In the core area of our practice — complex litigation — the answers are consistently strong; in areas where regulation is fragmented and changes frequently, performance is weaker, and that body of material is being improved separately. We cross-check the answers with our attorneys, but the final decision and responsibility always remain with the attorney. The tool does not replace the legal professional — it takes over the routine work of reviewing, organizing, and consolidating materials so that the lawyer can arrive at the ultimate result — a well-founded position on the case — sooner and with the full picture in view. That is why we are involved in its development.
