Upload a batch of case documents and get an index you can navigate: every document typed and dated, the parties named, the key dates pulled onto a chronology, and the whole set arranged as a map of issues rather than a folder of file names. Finding the right document stops being an afternoon.
The app embedded on this page is the Internal Docs App, a real Taskade Genesis build you can navigate right now. It organizes documents into a browsable structure, which is the closest thing here to a case index. Click "Use this app" to clone it in about ten seconds, then add the upload path, the reading step, and the party, date, and issue fields described below.
Concretely, a Matters Project holds one row per case and a Documents Project holds one row per file, kept as connected records so a matter with four hundred documents is still one matter. A published upload page takes the matter reference and a File field that accepts a batch. The Convert File to Text step reads each file, which matters here more than anywhere: disclosure sets are full of scanned letters, faxed pages, and photographs of documents, all sitting beside clean typed PDFs. An Ask Agent step returns document type, document date, author, recipients, named parties, a one-paragraph summary, and the issues it touches. A Categorize with AI step files it under your issue list. The chronology is assembled from the document dates, so the timeline is built from what the documents say rather than from someone's recollection.
The index reads in the shape the question takes:
- Mind Map view to see the matter as issues with their documents hanging off them
- Table view for the full document index, sortable by date and type
- Calendar view for the chronology laid out across real dates
- Plus List, Board, Gantt, and Org Chart across the 7 project views
Automations keep the index current. With Taskade automations, a newly uploaded batch is indexed and the team is notified of anything touching a hot issue, and a document dated inside a disputed window is flagged for a lawyer to read first. Across 100+ bidirectional integrations, forms and webhooks pull documents in while Slack and email push notices out. The AI agents inside the app carry 34 built-in tools including persistent memory and file analysis, so an agent can answer "which documents mention this term before March" against your own index.
Upload in batches and check each batch before starting the next. Two hundred documents indexed correctly is worth more than four hundred indexed hopefully, and the automation History tab is where you find out which it was. Expect the issue list to change as the matter develops, which is fine, because re-categorising is a step you can run again over documents already stored. Agree a naming convention for matters at the start so a second matter for the same client does not collide with the first. None of this is exotic, but on a document set of any size it is the difference between an index and a second pile.
Clone the app, invite the case team, and the index, the documents, and the issue list are yours.
See the reading step in Convert File to Text, the format list in Read Scanned Paperwork and Photos, and knowledge grounding in Agent Knowledge. Browse live builds in the Community Gallery, explore AI agents, or start from the templates library. Pair this with Build a Contract Register From Signed Agreements when contracts are in the bundle, and Capture Handwritten Notes as Real Action Items for attendance notes that belong to the same matter.
Frequently Asked Questions
Is the summary a substitute for reading the document?
No, and nobody on a matter should treat it as one. The index tells you which document to read and when it was written. The lawyer still reads it, and the original file is attached to every row for exactly that reason.
Can it handle a mixed bundle of scans and native files?
Yes, and that is the normal case. One Convert File to Text step reads a typed PDF, a scanned letter, a faxed page, and a photographed exhibit without branching by type.
How does the chronology get built?
From the document date field. Ask for the date written on the document rather than the file timestamp, then read the chronology in Calendar view. Documents with no date on their face are held in an undated group rather than being placed by guess.
Can I define my own issue list?
Yes. Provide the issues and the Categorize with AI step files each document against them. Documents matching more than one issue appear under each, since the views read the same underlying record.
Should confidential material go into a cloud workspace?
Check that against your own professional obligations before you upload anything. Taskade is a cloud service. Turn on app sign-in through App Users and restrict the workspace to the case team if you proceed.
How large can a batch be?
Each scanned file should stay under 20 MB, which is the ceiling for reading words off an image. Typed PDFs are read as text and bounded by your plan upload size. Very large bundles are best uploaded in batches so you can check each one in the automation History tab.
Can an agent answer questions about the set?
Yes. Feed the extracted text into Agent Knowledge and an agent answers grounded in your own documents, with the source row one click away.
Can we exclude privileged material from the index?
Decide that before uploading rather than after. Keep privileged documents out of the app entirely, or run them into a separate restricted matter with its own access, and record the exclusion in your own process notes rather than relying on a field in a shared workspace.
