> ## 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. # Reports — Complete Guide > Generate broker-quality, fully-cited research reports from reusable templates — what it does, when and how to use it, the templates and options, and how it fits your research workflow. Reports turns a few inputs into a finished, broker-quality research report. You pick or build a **template** — a reusable blueprint for a report — fill in a couple of fields (typically a ticker and a period), and the AI researches the company across filings, market data, analyst estimates, transcripts, news, and even your own documents, then writes a structured, multi-section report with charts, tables, and numbered citations on every sourced claim. You can watch it build live, edit it, keep a version history, and export it to a polished PDF or an editable Word document. Because every report runs from a template, you get the *same* professional structure every time — across companies, periods, and teammates. *Find it under **Research Workspace → Reports**.* ## When to use it Reach for Reports when: * You want a **formatted, multi-section deliverable** on a company or sector — an earnings review, an initiation, a valuation deep-dive, a peer comparison, a sector overview, an ESG or dividend write-up, a technical read, or a thematic note. * You want the **same structure produced consistently** across many names or periods — build the template once, then run it again and again with different inputs. * You need a **source-cited, professional document** you can share, archive, or hand to a client or committee. * You want the work **grounded in your own materials** as well as market data (your Data Room, uploaded files, or connected sources). Use something else when: * You just need a quick fact or a fast back-and-forth → **Chat**. * You have an open-ended question, or you want a **slide deck or an Excel model** as the output → **Agent Studio** (Reports exports to PDF and Word only — decks and models are an Agent Studio capability). * You want the same questions answered as a **structured table across many companies** → **Grids**. * You want to **find and gather source documents** first → **Document Search** and **Data Room**. A report is **always** generated from a template. There is no free-text "just write me a report" box — you choose a built-in template, build your own, or use one shared with you, fill in its inputs, and generate. This is what makes reports repeatable and consistent. ## How to use it Start from one of the **12 built-in AllMind templates**, one of **your own** saved templates, or a template **shared with you**. Opening a built-in template gives you your own editable copy — the original is never changed. Each template defines its own fields. Enter values using the right control for each — search for a **stock ticker**, add **several tickers** for a comparison, pick a **fiscal period** (e.g. Q4 FY2025), choose an **industry**, set a **date or date range**, or type free text. Required fields are marked; the **Generate** button stays disabled until they're filled. Click **Generate Report**. The report starts building on the server as a background job. You can launch **up to 3 reports at once** — a floating queue tracks them. The report streams in live: a research/data-gathering phase, then each section starting, the prose writing in, charts and tables appearing, and citations attaching as sources are found. A progress bar shows "Section X of Y." You can also follow a research-pipeline view for deeper reports. You don't have to wait — you can leave or close the tab and come back. When it's done, read the report with its charts and clickable citations. **Edit any section's text directly** (changes auto-save), **rename** it, or **ask the AI to revise it** in plain English and accept or discard its proposed changes. Every change is saved as a **version** you can compare and roll back to. **Export** the report to **PDF** or **Word**, or **share** it with a teammate or your whole organization. ### Building your own template You don't have to start from a built-in template. There are two ways to make your own: * **Create a template with AI** — describe the report you want in plain language ("an initiation report with business overview, financials, valuation, risks, and a recommendation"), or paste an existing report to mirror its structure. You can optionally **attach reference documents** (PDF, Word, Excel, CSV, or text, up to 12.5 MB each) so the AI models the template on your own material. It drafts the sections and input fields for you to refine. This runs as a quick background job (typically about a minute or two). * **Edit in the template editor** — open any template and adjust it directly: change the name and description, add or reorder **sections**, write each section's instruction, set its **output format** and **length**, and define the **input parameters** users fill in. Edits **auto-save** as you type. A template has three parts: | Part | What it is | | ------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------ | | **Details** | A name, a description, a writing **tone** (defaults to "professional"), and an optional overall length target. | | **Inputs** | The fields a user fills in each run (a ticker, a period, an industry, etc.) — each with a type, a label, and whether it's required. | | **Sections** | An ordered list of report sections. Each section has a heading, a plain-English instruction for what it should cover, an output format, and a target length. | Templates describe **what** each section should cover — they don't specify **where** to get the data. The report engine automatically chooses the right sources at generation time, so you can focus on the report's structure and let the AI find the numbers. ## Capabilities & key choices ### The 12 built-in AllMind templates Ready-to-run starting points, all finance-focused. Open one, fill its inputs, and generate — or save an edited copy as your own. | Template | Best for | | --------------------------------- | ----------------------------------------- | | **Earnings Review** | Recapping a company's latest quarter | | **Pre-Earnings Analysis** | Setting up for an upcoming print | | **Company Initiation** | A full initiation-of-coverage write-up | | **Valuation Deep Dive** | A focused valuation analysis | | **Technical Analysis** | Price action, levels, and indicators | | **Sector Overview** | Mapping a sector's landscape | | **ESG Analysis** | Sustainability and governance review | | **Dividend & Income Analysis** | Yield, payout, and income durability | | **Thematic Research** | A theme or trend across names | | **Analyst Consensus & Sentiment** | Estimates, targets, and the Street's view | | **Quick Stock Briefing** | A fast one-look summary on a name | | **Peer Comparison** | Comparing several companies side by side | ### Per-section choices When you build or edit a template, each section can be tuned: * **Output format** — **Markdown (narrative text)** for flowing analysis, **Table** for structured rows and columns, or **Bullet Points** for quick-scan lists. * **Target length** — **Brief (\~300 words)**, **Moderate (\~600 words)**, **Comprehensive (\~1200 words)**, or an exact custom word count. You can also set an overall length target for the whole report. * **Order and dependencies** — sections are written in order, so each one can build on the ones before it, which keeps the report consistent and avoids repetition. ### Input parameter types Each input renders the right control for its type: | Type | What the user does | | ------------------------- | -------------------------------------------- | | **Stock Ticker** | Search a single company by symbol or name | | **Multiple Stocks** | Add several tickers for a comparison | | **Fiscal Period** | Pick a quarter or full year (e.g. Q4 FY2025) | | **Industry** | Enter an industry or sector | | **Date** / **Date Range** | Choose a date or a start/end range | | **Number** | Enter a numeric value | | **Text** / **List** | Free text, or a comma-separated list | ### Watching generation For richer reports you can follow the work as it happens: * A **section-by-section progress bar** and a live **research-activity feed** showing what the AI is doing (pulling filings, fetching consensus metrics, generating a chart, reasoning). * An optional **research-pipeline view** that progresses through planning, searching, and synthesis stages, with an indication of how much research effort the task warrants (from a fast single pass to a deep multi-step run). * An automatic **fact-check on earnings reports** that verifies the reported figures (EPS, revenue, and margins) against the underlying filings and consensus. Reports can also be **scored for quality** — checked for repetition, numerical consistency, and contradictions, and given a score and grade. ### Editing, versions, and revising with AI A finished report is a living document, not a static read-out: * **Inline editing** — click any paragraph or table to edit it in place; changes **auto-save** about a second and a half after you stop typing. * **Edit with AI** — ask for changes in plain English ("tighten the risks section," "add a competitive landscape section," "update every FY24 reference to FY25"). The AI proposes revisions shown as a **red/green diff**; you **Accept** or **Discard**, and any new sources it cites are folded into the report. * **Version history** — every report keeps a history (created on generation, on edits, on accepted AI changes, on rename, and on restore). Open any past version, see exactly what changed, and **restore** it — restoring itself creates a new version, so you never lose work. ### Export & sharing * **Export** to a print-ready **PDF** or a fully editable **Word (.docx)** document. Both are styled like a sell-side research note — branded masthead, the report title and ticker, headings, tables, embedded charts as numbered figures, **clickable citations**, a numbered **References** section, and a standard informational disclaimer. (A plain **Markdown** export is also available.) * **Share** a report or a template with a single teammate or your **whole organization**, at **read**, **write**, or **admin** level — with independent toggles for whether they can **download**, **export**, and **re-share**, and an optional **expiry date**. Items shared with you appear marked "Shared." A shared **template** can be copied into your own library and edited; reports can be shared but not copied this way. ## What it can access & produce **It can draw on:** * **Company filings** — US (SEC) and Canadian (SEDAR) filings and other primary documents. * **Financials** — financial statements, fundamentals, ratios, segments, and adjusted metrics. * **Market data** — live and historical prices for stocks, indices, and commodities; technical indicators; and sentiment and institutional-flow signals (including dark-pool activity). * **Estimates & the Street** — analyst consensus, estimates, price targets, and broker recommendations. * **Transcripts & research** — earnings-call and investor-relations transcripts, and broker/analyst research. * **News, ESG, and macro** — company and market news, ESG ratings, and macro and economic indicators. * **The web.** * **Your own materials (optional)** — your Data Room and uploaded documents, plus connected sources like Google Drive and a connected organizational database, when enabled. It also takes the **template's inputs** (the ticker, period, etc.) and any **reference files** you attach when creating a template with AI. **It produces:** * An on-screen, **multi-section report** with narrative, tables, charts, and numbered inline citations linked to a sources panel. * A **PDF**, a **Word document**, or a **Markdown** file. * A **saved, versioned** report that persists in your history. * A **reusable template** (when you build or learn one) that can feed future reports. ## Tips & best practices * **Let the built-in templates do the heavy lifting** — they cover the most common report types and give you a professional structure instantly. Save an edited copy to make one your own. * **Be specific in section instructions** when you build a template — the clearer you are about what a section should cover, the better the result, every time you run it. * **Use per-section length and format deliberately** — a Table for a peer comparison, Bullet Points for a quick-take summary, Comprehensive narrative for the analysis that matters. * **Ground it in your own documents** by enabling your private data or attaching reference files when the analysis should reflect your materials, not just public data. * **Run several at once** — launch up to three reports in parallel (e.g. the same template across three peers) and let them build while you work. * **Refine with the AI rather than starting over** — ask for targeted revisions and accept the diffs; every change is versioned, so you can always roll back. ## Limits & things to know * **Always template-driven** — you generate from a template, never a blank prompt. * **Up to 3 reports** can generate at the same time. * **It runs in the background** — generation can take from a few minutes to tens of minutes depending on the number of sections, their length, and how much research is needed. You can close the tab and come back; the report keeps building and even reuses the research it already gathered if you resume. Very long runs are capped (each section has roughly 10 minutes, and a whole report up to about 2 hours, before it times out). * **Reference files** for AI template creation must be **PDF, Word, Excel, CSV, or text, up to 12.5 MB each.** * **Export formats are PDF and Word only** (plus Markdown). There is **no PowerPoint deck or Excel model** export from Reports — those are an **Agent Studio** capability. * **Reports and templates persist** until you delete them, and deleting is **permanent**. * **Reports are finance-focused** — every template is built for financial analysis. * **Reports are heavily sourced** — a typical multi-section report aims for at least roughly 15 distinct cited sources, with citations throughout. ## How it works with other features Reports is the **deliverable** stage of the research workflow, and it connects across the platform: * **Document Search → Data Room → Reports** — find the right sources, collect them in a Data Room, then let a report draw on (and cite) your own documents alongside market data. * **Agent Studio ↔ Reports** — an agent run is best for open-ended research and for producing decks or models; a report is best when you want a consistent, templated, formatted document. Use an agent to explore, then a report to package the finished analysis. * **Grids ↔ Reports** — a Grid answers the same questions consistently across many companies as a table; a report is the narrative, written-up complement. Use a Grid to compare, a report to explain. * **Chat → Reports** — the same conversational assistant that powers Chat is what you use to revise a report in plain English. * **Automations → Reports** — a report template can be run on a schedule (for example, generate an earnings review automatically when a company reports), with the finished PDF or Word file emailed to recipients. **Inputs that feed a report:** a template, its input values, your own documents, and connected data. **Outputs a report hands off:** a cited PDF/Word/Markdown document, a saved report, and reusable templates — ready to share, archive, or attach to the next step. ## Example workflows ### Build a repeatable earnings-review process 1. Open the built-in **Earnings Review** template and tweak its sections to match your house style (your preferred KPIs, a guidance section, a "what changed vs. last quarter" section). Save it as your own template. 2. Each quarter, **generate** the report for a name by entering its ticker and the new fiscal period. 3. Let the AI pull the latest filing, transcript, consensus, and reaction data and write the report, with the extra **earnings fact-check** verifying the headline numbers. 4. **Refine** any section with the AI, then **export to PDF** and **share** it with your team — or set an **Automation** to run it the moment the company reports. ### Analyze a sector and produce a presentation-ready set A multi-feature journey — "analyze the gold sector and brief the team": 1. **Scope it** — list the key names and the 7–8 topics that matter (cost curves, reserves, macro drivers, hedging, capital returns, risks). 2. **Gather sources** — use **Document Search** to find the relevant filings, transcripts, and research, and collect the best into a **Data Room**. 3. **Research the names** — fire several **Agent Studio** runs, one per sub-topic, for deep narrative analysis. 4. **Compare consistently** — build a **Grid** to answer the same questions across the producers. 5. **Write it up** — run the **Sector Overview** report template for the sector view and the **Peer Comparison** template across the producers, grounded in your Data Room. 6. **Deliver** — export the reports to **PDF/Word** to share, and (if you want slides) hand the findings to **Agent Studio** to build the deck. ## Common questions Reports are always generated from a template — that's what makes them consistent and repeatable. But making one is easy: start from a built-in AllMind template, or describe the report you want in plain language and let the AI draft the template's sections and inputs for you. Twelve finance templates: Earnings Review, Pre-Earnings Analysis, Company Initiation, Valuation Deep Dive, Technical Analysis, Sector Overview, ESG Analysis, Dividend & Income Analysis, Thematic Research, Analyst Consensus & Sentiment, Quick Stock Briefing, and Peer Comparison. Opening one gives you your own editable copy. Company filings (US and Canadian), financial statements and fundamentals, live and historical prices and technical indicators, analyst estimates and consensus, price targets and broker recommendations, earnings-call and investor-relations transcripts, broker research, news, ESG ratings, macro indicators, sentiment and institutional-flow signals, and the web — plus your own documents and connected sources when you enable them. Every sourced claim gets a numbered, clickable citation. It runs in the background and can take from a few minutes to tens of minutes depending on the number and length of sections and how much research is needed. You can close the tab or navigate away — the report keeps building and is waiting for you when you return; if you reopen it mid-run, it resumes and reuses the research it already did. Yes — up to three at once. A floating queue tracks each one's progress, and you can view, cancel, or clear them from there. Yes. Click any paragraph or table to edit it in place (changes auto-save), or ask the AI to revise a section or the whole report in plain English and accept or discard its proposed changes. You can also rename the report and its sections. Yes. Every report keeps a version history — created when it's generated, when you edit, when you accept AI changes, when you rename, and when you restore. You can open any past version, see a red/green diff of what changed, and restore it. Restoring creates a new version, so nothing is lost. PDF and editable Word (.docx), plus Markdown. Both PDF and Word are styled like a research note — branded header, headings, tables, embedded charts as numbered figures, clickable citations, a References section, and a disclaimer. There is no PowerPoint or Excel export from Reports — those are an Agent Studio capability. Earnings reports get an automatic fact-check that verifies the headline figures (EPS, revenue, and margins) against the underlying filings and consensus. Reports can also be scored for quality — checked for repetition, numerical consistency, and contradictions — and given a score and grade. Yes. Share a report or template with one teammate or your whole organization, at read, write, or admin level, with independent controls over downloading, exporting, and re-sharing, and an optional expiry date. Anything shared with you appears marked "Shared," and a shared template can be copied into your own library. Yes — when enabled, a report can search and cite your Data Room and uploaded documents, and connected sources like Google Drive and a connected organizational database, alongside public market data. You can also attach reference files when creating a template with AI. ## Getting help For help using Reports or to talk through a reporting workflow, reach AllMind support through the in-app support option or your account team.