> ## Documentation Index
> Fetch the complete documentation index at: https://complete-docs-x17s80.allmind.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Data Room — Complete Guide
> Your private, AI-ready research libraries — collect filings, transcripts, news, uploads, and authorized warehouse data into shareable folders, then ask cited questions across them at any scale.
A **Data Room** is a private collection of source material that the AI treats as one focused universe. Each room is a folder, and a folder can hold company filings, earnings-call and event transcripts, news, data tables, your own uploaded files, AI-generated insights you've saved, and authorized data-warehouse tables. Once a room is populated, you can ask questions across it and get answers that **cite the exact source documents** they came from, share the room with colleagues, and point other AllMind tools at it. Its defining strength is **scale and grounding**: the same workflow works whether a room holds a dozen documents or tens of thousands, and the answers are anchored in *your* materials rather than the open web.
*Find it under **Research Workspace → Data Room** — it collects your folders, saved documents, uploads, and files.*
## When to use it
Reach for a Data Room when:
* You're running a project — a company deep-dive, a sector study, a deal, a watchlist — and want all its sources in one place.
* You want the AI to answer **only from sources you trust**, with citations you can open and check.
* You want to gather a large body of filings or transcripts and ask questions across the whole set without reading each file.
* You want your own files (or your firm's warehouse data) to be part of the analysis.
* You want a curated source set you can hand to a colleague, or reuse to ground Chat, an agent run, or a report.
Use something else when:
* You just want to **find** documents in the first place → start in **Document Search**, then save the good ones into a Data Room.
* You want a quick, one-off answer with no saved sources → **Chat**.
* You want a thorough, autonomous multi-step analysis → **Agent Studio** (you can point it at a Data Room).
* You want the same questions answered consistently across many companies or documents in a table → **Grids**.
* You want a formatted, templated deliverable as the output → **Reports** (a Data Room can be its source).
## How to use it
Each folder is its own research room. Give it a name and an optional description. Folders are **flat** — there are no sub-folders — and they all appear together in your folder list alongside an **All Data** view and any folders shared with you.
Populate the room in whichever way fits — search AllMind's document library, upload your own files, add authorized warehouse tables, save items from elsewhere in the platform, or let AI assemble it for you. (Each path is covered under *Capabilities* below.)
Switch between a grid or list view. Search by company, title, or type (the search is forgiving — “att” finds “AT\&T”, “6k” finds “6-K”, “aple” finds “Apple”). Narrow the view by content type, by document type, by company, or by a recent time window (today, last 3 / 5 / 10 / 30 days), and sort by document date, date added, or title. Open any document to read it inside the room.
Use **Ask AI** to start a chat grounded in the folder. Ask in plain language and get an answer with **numbered citations** that link back to the exact source documents — click any citation to open the source. You can ask about a whole folder, all your folders, or just the items you've selected.
Turn on **Auto-Update** so new matching filings flow into the folder automatically, and **Share** the folder with colleagues or your whole organization when you want others to use it.
Everything you save is stored to your AllMind account automatically, so your rooms follow you to any device you sign in on — no manual saving or syncing.
## Capabilities & key choices
### Ways to fill a room
The in-room library search. Choose the destination folder, add one or more companies, then pick what to pull:
* **Templates** — one-click presets that pre-set the years, categories, and document types for a use case (for example *Earnings Analysis – 5 Years*, *Annual Reports*, *M\&A / Special Situations*, *Mining & Energy technical reports*, *Capital Markets & Offerings*, *Insider Trading & Ownership*).
* **Year range** — last 2 / 5 / 10 years, all, or a custom from/to (defaults to the last 5 years).
* **Document categories** — **SEC filings** (US — annual reports/10-K, quarterly/10-Q, current/8-K, proxies, registrations, prospectuses, insider/ownership, and more), **SEDAR filings** (Canada — annual reports, financial statements, MD\&A, annual information forms, material-change reports, prospectuses, NI 43-101 technical reports, NI 51-101 oil & gas reports, take-over bids, ownership filings), **earnings-call transcripts**, **earnings-call slides**, **press releases**, **third-party research reports** (up to a two-year window), **ESG reports**, and **investor-relations events** (non-earnings conference talks, investor days, shareholder meetings — searchable by company or across companies for a date range).
* **Keyword search** — free-text search across document types with a result-count limit, for when you're hunting a specific topic rather than a filing type.
Results stream in pre-selected; review them, deselect anything you don't want, then add the rest to the folder.
Describe the research set you need in plain language — you can even dictate it by voice — for example *“Palantir 10-Ks, 10-Qs and earnings transcripts for the last two years, plus the semiconductor sector.”* AllMind proposes a full **plan**: which companies (it resolves sectors and ETFs/indices into their constituent companies, and flags any name it couldn't place), which document types, and which years and quarters. You **review and edit** the plan — add or drop companies, adjust the period, toggle document types, or pick individual documents — and **nothing is added until you approve it.**
Add your own documents by dragging them in or browsing your computer, or import them from **Google Drive, OneDrive, SharePoint, Gmail, or Outlook** once those accounts are connected (connections you've already set up in chat carry over). From Gmail and Outlook you bring in emails as documents, with attachments optional. You can add up to **150 files per upload, 100 MB each**, in common formats — PDF, Word, Excel, PowerPoint, CSV, text, HTML, Markdown, and OpenDocument spreadsheets.
If your organization has connected its Snowflake data warehouse, you can scope **authorized tables** into a folder so the AI can query that data alongside your documents in the same chat. You can only pick from the tables your organization has approved for you — there's no free-text entry. Access is **read-only**, and if a table's access is later revoked it's marked *Access revoked* in your room and the AI automatically skips it.
Use the save/bookmark control wherever you find something useful and drop it into a folder: **documents** from Document Search (one at a time or in bulk), **news** articles, **data tables** from the company data views, **earnings-call** and **event transcripts**, and **AI-generated insights** you've produced. You choose the destination folder as you save.
### Asking the room questions
* **Scope** — ask about one folder, all folders, or only the items you've selected. Started from inside the Data Room, the AI is automatically pointed at your documents and told to **search them first**, before the web or other libraries.
* **Data-Room-only mode** — you can restrict the assistant so it answers **purely from your Data Room** (and its authorized tables), turning off the web and other sources for that conversation.
* **Citations** — every fact the AI draws from a document is tagged with a numbered citation that opens the exact source. Sources that came from a warehouse table are shown distinctly and open a row preview.
* **Automatic scaling** — small rooms have their documents loaded straight into the conversation; larger rooms switch to a mode where the AI works through search, so even very large libraries stay responsive.
## What it can access & produce
**A room can draw on:**
* Company filings — US (SEC) and Canada (SEDAR).
* Earnings-call transcripts and slides, and investor-relations event transcripts.
* Press releases, third-party research, and ESG reports.
* News articles and saved company data tables.
* Your own uploaded files (and files imported from Google Drive, OneDrive, SharePoint, Gmail, and Outlook).
* AI-generated insights you've saved.
* Authorized data-warehouse (Snowflake) tables, where your organization has connected them.
**It produces:** a curated, searchable, shareable source set — and, through Ask AI, **cited answers** grounded in those exact sources. A room is also an **input** other features can build on (see below).
## Tips & best practices
* **One room per project.** Keep a company, sector, deal, or theme in its own folder so the AI's answers stay focused and your citations stay relevant.
* **Let AI do the assembly, then refine.** “Build Data Room with AI” is the fastest way to stand up a room; tune the proposed plan before approving rather than adding everything by hand.
* **Use Auto-Update for living coverage.** For names you track, set a subscription so new filings arrive on their own — but add any *older* documents yourself, since Auto-Update only brings in new ones going forward.
* **Restrict to the room when grounding matters.** Turn on Data-Room-only mode when you need answers strictly from your trusted sources, with no web influence.
* **Ask focused questions for the best citations.** Meaning-based search returns the most relevant sources, which is ideal for analysis and summaries. For an *exhaustive* “list every document from this week” task, lean on connected warehouse data, which is built for complete pulls.
* **Select before you ask.** When you only care about a few documents, select them and ask AI about just those.
## Limits & things to know
* **Folders are flat** — no sub-folders or nesting. Organize with multiple folders and clear names.
* **Moving items between folders isn't available right now.** You choose a folder when you add a document, and you can remove a document from a folder; to relocate a library document, add it again to the other folder. (Each saved item lives in one folder; the same warehouse table can be added to more than one.)
* **Auto-Update is forward-only.** It adds newly published matching documents from the moment you set it up — it does not backfill documents that already existed.
* **Ask AI surfaces the most relevant sources, not all of them.** Any single answer is grounded in the top matches for your question, so it isn't a guarantee that every matching document was read. For complete coverage of a large set, use connected warehouse data.
* **Warehouse (Snowflake) data is organization-gated and read-only.** It appears only if your organization has connected it and authorized specific tables; revoked tables are skipped automatically, and tables authorized for one organization aren't used by someone from another organization who opens a shared folder.
* **AI-assisted setup depends on your workspace.** If it isn't enabled for your workspace yet, you'll see a message to that effect; you can still build rooms manually.
* **Deleting a folder removes everything in it** — its documents and its Auto-Update subscriptions — so delete deliberately.
* **Your rooms persist** until you delete them; nothing expires on its own, and everything is saved to your account and available across devices.
## How it works with other features
A Data Room sits at the center of the research workflow — most features feed into it or build on it.
Find filings, transcripts, and research, then save the best results — individually or in bulk — into a folder.
Ask cited questions grounded in a folder, optionally restricted to your sources only.
Attach a folder (its documents and authorized tables) to an agent run so autonomous research is grounded in your materials.
Generate a report scoped to a folder, so the report is built from — and cites — that source set.
Grids work a little differently: a Grid analyzes the documents you attach **to that Grid**, not your existing Data Room folders — though adding documents to a Grid works just like building a Data Room.
## Example workflow: a grounded sector study and presentation
A real multi-feature journey — *“analyze the gold sector and build a presentation”* — with the Data Room as the source of truth:
1. **Stand up the room** — open the Data Room and use **Build Data Room with AI**: describe the set (“major gold producers — 10-Ks, earnings transcripts, and recent research, last 2 years”), let it resolve the sector into companies, review the plan, and approve.
2. **Fill the gaps** — use **Document Search** to find any specific notes or filings the plan missed and **save** them into the same folder; **upload** your own model or memo if you have one.
3. **Keep it live** — turn on **Auto-Update** for the key tickers so new filings keep arriving while you work.
4. **Interrogate it** — use **Ask AI** (Data-Room-only) to pull out supply/demand drivers, cost curves, and risks, checking the **citations** as you go.
5. **Go deep where it matters** — point an **Agent Studio** run at the folder for the meatier sub-questions.
6. **Structure it** — build a **Grid** over the producers to answer the same questions consistently across companies.
7. **Deliver** — generate a **Report** scoped to the folder, so the final deck or document is grounded in — and cites — the exact sources you assembled.
The result: a sourced, organized research base, reused end-to-end, with a finished deliverable that traces back to your documents.
## Common questions
Document Search is how you *find* sources across the platform; a Data Room is where you *keep and use* the ones you choose — as a focused, citable set you can ask questions across, share, and feed to other tools.
When you start from a Data Room, the AI is pointed at that folder and told to search your documents first. You can also restrict it to answer only from your Data Room, with the web and other sources turned off.
Yes — answers carry numbered citations, and each one opens the exact document (or warehouse row) it came from, so you can verify every claim.
Yes. Large rooms switch automatically to a search-driven mode, so the AI stays aware of the whole set and pulls the most relevant sources for each question. For a truly exhaustive “list everything” pull, use connected warehouse data.
Yes — upload from your computer or import from Google Drive, OneDrive, SharePoint, Gmail, or Outlook (up to 150 files per upload, 100 MB each, in common document and spreadsheet formats).
Not directly right now. You set the folder when you add a document, and you can remove it; to relocate a library document, add it again to the other folder.
No — it only adds new matching documents from when you set it up. Add any earlier documents yourself.
Yes — share a folder with specific colleagues or your whole organization, choose View, Edit, or Admin access, optionally set it to expire, and control whether recipients can download, export, or re-share it. You can see who has access and change or revoke it anytime.
Everything in it is removed — its documents and its Auto-Update subscriptions — so delete with care.
No — it's part of the platform. Connecting your organization's data warehouse (Snowflake) is the one capability that depends on an organization setting up that connection.
## Getting help
For help building a Data Room or talking through a workflow, reach AllMind support through the in-app support option or your account team.