Drafting is where discovery work actually gets used. The motion, the deposition outline, the privilege log narrative — all of it is written in Word, while the evidence sits in another window. So the work becomes a loop of alt-tabbing: find the document, copy the quote, check the Bates number, come back, lose the thread. The add-in closes that loop. Your matter opens as a task pane beside the document, with the same permissions and the same record you already work from.
Four things, without leaving the page
Draft and revise with case context
Generate a passage or rework one you have written, grounded in the documents already loaded into the matter rather than in a model with no knowledge of your case.
Why it matters. A general-purpose assistant will write you a fluent paragraph about a deposition it has never read. This one works from the record you produced.
Cite and pull from case documents
Search the matter's document set from the task pane, read the hit in context, and drop the excerpt into your draft with its source attached.
Why it matters. The quote and the citation travel together, so the thing you paste into a brief is traceable back to the document it came from.
Privilege log and redaction assist
Flag privileged or confidential passages as you write, and build privilege log entries from the material in front of you instead of reconstructing them later.
Why it matters. Privilege calls made at the moment of drafting are better calls, and they leave the log entry behind as a by-product.
Summarize and analyze the open document
Summarize the document on screen, pull out the facts that matter, and place events onto the matter's chronology without re-uploading anything.
Why it matters. The draft in front of you is evidence too — a transcript, a contract, a production letter. Analyzing it in place saves a round trip.
The same matter, on both sides of the screen
The add-in is not a separate product with its own data. It is a second surface onto the matter you already run in DecoverAI — the same documents, the same classifications, the same access rules.
Load the matter in DecoverAI
Documents are processed, classified for responsiveness, privilege, and confidentiality, and made searchable in the web app as they are today.
Open the task pane in Word
Sign in with your DecoverAI account and pick the matter. Your matter permissions carry over exactly — the add-in cannot show you anything the web app would not.
Draft against the record
Search, cite, summarize, and flag as you write. Work you do in the pane stays attached to the matter, so nothing is stranded in a local document.
Where it runs, and what you need
Supported versions of Word
- Word for Microsoft 365 on Windows and macOS
- Word on the web, in any modern browser
- Word 2021 and Word 2019, signed in with a Microsoft 365 account
- A signed-in Microsoft 365 account, which Office requires to load any add-in
What your firm needs
- A DecoverAI account — the add-in is included, with nothing separate to license
- Outbound HTTPS to
*.decover.aion port 443, if your firm filters traffic - Optional: central deployment by your Microsoft 365 administrator through Integrated apps
- No local install, no MSI, and no desktop agent — it loads from the Office Store
Held to the same standard as the platform
Your documents are not training data
The add-in inherits every control that governs DecoverAI itself: AES-256 encryption at rest, TLS 1.2+ in transit, role-based access, full audit logging, and matter-level permissions carried across from the web app. Nothing you write or open in Word is used to train, fine-tune, or improve AI models.
Document content reaches DecoverAI only when you ask for an action, and is processed in isolated environments that are destroyed when the request completes. Details are on our security page, in our privacy policy and data processing agreement, and in the Trust Center, where SOC 2 Type II and HIPAA documentation is available.
Installing it, or stuck on something?
Step-by-step install instructions, admin deployment notes, and troubleshooting for the add-in live on our support page.