Divorce Masters is a legal technology platform for attorneys and their teams working through financial disclosure. The job spans clients, requested records, missing information, transaction review, professional judgment, and reporting. A useful system has to keep those pieces connected to the same matter.
33Labs' consultancy covered product architecture, the firm workspace, a client document portal, backend systems, and AI-assisted financial review. The goal was not to place a chatbot over an existing process. It was to build software around the actual handoffs between clients, paralegals, administrators, and attorneys.
Financial disclosure is not complete when files appear in a shared folder. The legal team needs to know what was requested, what arrived, what is still missing, and which findings require a human decision.
Clients and staff approach that work with very different context. A client may be on a phone, under stress, and using the portal for the first time. A paralegal needs to see the state of many requests. An attorney needs to inspect the financial picture and decide what the information means.
The workflow also contains distinctions that generic document processing misses. A missing upload does not necessarily mean the client forgot a file; the requested record may not exist. A transaction labeled as a utility does not tell the team whether it belongs under electricity, water, telephone, or gas on a financial affidavit.
Those are product and data-model decisions, not just prompt-writing problems.
Clients upload against a checklist organized around Florida Rule 12.285 mandatory disclosure. A separate attestation lets a client state that a requested document does not exist, giving the legal team an explicit status instead of an unexplained empty slot.
Attorneys and staff organize work around the matter. The checklist, incoming documents, AI-flagged issues, and financial corrections stay connected to the case they belong to. The client upload process therefore has a destination inside the firm's working file rather than ending at receipt.
The transaction ledger supports filtering and paging so staff can work through targeted review queues. AI-proposed categories remain distinct from staff corrections and affidavit-line assignments.
Firm-wide merchant rules can provide reusable defaults, while a rule specific to one matter takes priority when that case needs different treatment. The reusable element is the firm's categorization convention—not one client's financial records being transferred into another client's matter.
The reporting backend carries detailed affidavit lines into the workbook alongside broader spending summaries. Those views remain separate so the same dollar is not counted twice. Staff can inspect both the high-level financial picture and the specific reporting line without turning either view into a duplicate transaction.
The application uses Next.js for the firm workspace and client portal, with a NestJS and Prisma backend and Supabase authentication. Tenant-data requests go through the backend with the caller's own token; the frontend does not hold service-role credentials.
The architecture keeps AI assistance separate from legal judgment. The system may propose a category or surface an issue, but the legal team can inspect, correct, and approve the information before relying on it. That boundary is central to the product: assistance should make the working file easier to assemble and review without presenting model output as a finished legal conclusion.
The July 2026 implementation established the branded workspace, client portal, and disclosure-checklist flow. August delivery added filtered transaction review and finer affidavit-line reporting.
The practical result is a clearer place for each participant to act:
This is how 33Labs approaches legal AI product consultancy: understand the professional workflow, encode its important distinctions, use AI where it improves review, and preserve human control where judgment carries consequences.
This case study describes implementation milestones, not measured time savings, accuracy gains, adoption figures, or the production status of any specific law firm.