The platform
A legal operating system, not a pile of tools.
Most firms run a research subscription, a practice management system, a document store, an accounts package and an AI tab, and spend their day carrying information between them. ModulawAI puts research, drafting, review, matters, money, clients and agents on one data model, so the assistant that drafts the letter already knows the matter it belongs to and the clock that should be running. This page lists everything in it.
Definition
What a legal operating system is.
A legal operating system is a single platform that holds both the knowledge work of a law firm and the running of the firm itself, on one data model, so that the two can see each other. Research, drafting and document review sit in the same system as matters, deadlines, time, billing, clients and staff.
That distinction is the whole argument. An AI research tool can find the authority but cannot open the matter, record the time or raise the invoice. A practice management system can do all three but has no idea what the law says. When the two are separate products, a person is the integration, and that person is a lawyer doing data entry.
On one data model the work joins up. The agent drafting a letter knows which matter it belongs to and which client is billed. The timer knows what was being worked on. The document knows which file it lives in and who is allowed to see it. Nothing has to be re-entered for the next tool to understand it.
Layer one
Research and analysis.
Everything here retrieves the source before it answers, and links what it cites. Coverage spans Nigeria, the United States, the United Kingdom and Ireland, Canada, the European Union, Ghana, Kenya and South Africa.
| Tool | What it does |
|---|---|
| Case law research | Natural-language search across eight jurisdictions, with the authority, its treatment and a link you can open |
| Statutes, regulations and court rules | Search the Nigerian corpus including Acts, regulations and the civil procedure rules of each court |
| Citation verification | Batch-check that every citation in a piece of research actually exists at the source it claims |
| Contradiction checker | Systematically hunts for authority that cuts against your position before the other side finds it |
| Deep analysis | A long-running multi-source analysis for questions that will not yield to a single search |
| Citation graph analytics | Citation networks, judicial patterns and court behaviour, queried over a legal knowledge graph |
| Legal Library | Deterministic browsing of the Nigerian corpus when you want to read rather than ask |
| Web research | Live web search and a visual browsing agent for the things that are not in any law report |
Layer two
Drafting and documents.
Drafting happens either in the Canvas, a long-form editor beside the chat, or in Microsoft Word where most lawyers actually work. Citations survive the journey in both directions.
| Tool | What it does |
|---|---|
| Long-form legal drafting | Written addresses, briefs, opinions, memoranda, contracts, formal letters and case reports, opened in the Canvas as a real document |
| Microsoft Word add-in | Native track-changes redlines and comments anchored to the paragraph, inside the document rather than beside it |
| Document templates | A seeded template library built as structured blocks and exported to .docx, with Word content controls for the fields |
| Precedent library | A workspace document store the assistant can draft from, so new work starts from your own precedents |
| Reports | Formal reports and memos exported as PDF or Word |
| Designed reports | Magazine-quality PDFs with real typography, charts and editorial layout for anything a client will see |
| Slide decks | Native PowerPoint .pptx files generated from the work |
| Spreadsheets | Native Excel .xlsx workbooks and CSV exports, styled rather than dumped |
| Visuals and imagery | Designed cards, posters and infographics, plus generated imagery where a document needs one |
Layer three
Review at scale.
Reading one document is a chat. Reading two hundred is a different problem, and it needs a grid rather than a conversation.
| Tool | What it does |
|---|---|
| Tabular Review | A spreadsheet over a set of documents. Every column is a question asked of every document, and every cell carries the page it came from |
| Document processing | Reads PDFs, Word files and scans, extracts the text and caches it so the same pages are not paid for twice |
| Issue and citation extraction | Pulls the issues, the authorities cited and the statutory cross-references out of a document you upload |
Layer four
Running the firm.
This is the half that AI-only tools do not have, and it is the half that decides whether the AI is useful or merely impressive.
| Area | What it covers |
|---|---|
| Matters and cases | The file itself: parties, status, deadlines, documents, notes and everything filed against it |
| Tasks and calendar | Work assigned to people, with the court dates and limitation deadlines that cannot slip |
| Notes | Attendance notes and file notes, attached to the matter rather than a notebook |
| Time and billing | Capture from the work itself through to an invoice, in the currency the client pays in |
| Financials | Invoicing, payments, and the reporting a partner asks for at month end |
| Requisitions and purchase orders | Multi-line internal spend requests routed through an approval chain, scoped to a case where relevant |
| Approvals | A shared approval engine with workspace policies and per-case rules |
| Workspaces, seats and practice groups | Firm structure, teams, roles and who can see which matters |
| Analytics | Reporting across matters, people, time and money |
Layer five
Clients, intake and communication.
Everything the client touches, and every channel the firm answers on, recorded against the matter rather than scattered across personal inboxes.
| Area | What it covers |
|---|---|
| Client portal | A place for the client to see progress, documents and invoices without an email chain |
| CRM and pipeline | Prospects, custom fields, lifecycle stages and the work of turning an enquiry into a matter |
| Intake and forms | Structured intake, conflict-relevant detail captured once, and forms that feed the file directly |
| Bookings and scheduling | Consultation booking, including public guest booking pages |
| A real mailbox integration over IMAP and SMTP, plus Zoho Mail, with the inbox attached to the practice | |
| Team chat | Internal channels where the assistant can be summoned into the conversation |
| Voice | A voice assistant you can talk to, and voice calls handled with transcription |
| The channel most clients in our core markets actually use |
Layer six
Agents, automation and building your own.
The part most legal software does not have at all: the ability to make the system do something new without waiting for a vendor release.
| Capability | What it does |
|---|---|
| Workflows and processes | Repeatable matter processes with steps, owners, forms and gates |
| Autonomous agents | Long-running agents that carry out multi-step work and report back, rather than answering one message |
| Canvas apps | Describe a tool and get a working, interactive application rendered live in the Canvas |
| Code interpreter | Python, JavaScript and TypeScript executed in an ephemeral sandbox for calculations and data work |
| Skills | Packaged instructions, reference files and scripts that teach the assistant your firm’s way of doing a task |
| Interactive panels | The assistant renders real interface components beside the chat to show structured data or collect input |
| REST API | A scoped API for your own systems to read and write |
| MCP server | Connect Claude, ChatGPT or any MCP client to your firm’s matters, documents and research |
| External MCP servers | Connect third-party MCP servers in the other direction, so the assistant can reach your other tools |
Surfaces
Where the system runs.
| Surface | What it is for |
|---|---|
| Browser | The full system: matters, documents, research, financials and the assistant |
| Microsoft Word | Drafting and redlining where the document already lives |
| Desktop, Mac | The same system with offline access, syncing when the connection returns. A Windows version is in progress |
| Voice | Asking and answering without a screen |
| Team chat and WhatsApp | Where conversations already happen, internally and with clients |
| API and MCP | Where your own software, or another AI client, plugs in |
FAQ
Questions, addressed.
What is a legal operating system?
A legal operating system is a single platform that holds both the knowledge work of a law firm and the running of the firm, on one data model. Research, drafting and document review sit alongside matters, deadlines, time recording, billing, clients and staff, so the AI can see the practice and the practice can see the AI. The alternative, which most firms run today, is a research subscription and a practice management system that know nothing about each other.
How is this different from practice management software with AI added?
Order of construction. A practice management system with an AI feature bolted on can summarise what is already in the database. A system built AI-first retrieves primary law, drafts from your precedents, reviews document sets and runs agents, then writes the results back into the matter. The test is simple: ask whether the AI can cite a judgment you can open, and whether the answer it gives knows which client is being billed for it.
Do we have to move everything at once?
No. There is a scoped REST API and an MCP server, so ModulawAI can sit next to what you already run and be reached from it. Several firms run it alongside an existing research subscription for a period before deciding. Matters, documents, time entries and invoices export in standard formats if you decide against it.
Which jurisdictions does the research cover?
Nigeria, the United States, the United Kingdom and Ireland, Canada, the European Union, Ghana, Kenya and South Africa. Comparative authorities surface across borders where useful.
Can we build our own tools on it?
Yes, three ways. Canvas turns a description into a working interactive application. Skills package your firm’s own instructions, reference files and scripts so the assistant follows your method. And the REST API and MCP server let your own systems, or another AI client, read and write directly.
What does it cost?
Plans are published rather than quote-only and billed in your local currency, with a monthly pool of AI credits shared across research, drafting, document reading, voice and agents. See pricing for current figures.
Keep reading
Go deeper on any layer.
See the whole thing on one matter.
Book a demo and bring a live file. We will run it from research through to the invoice.