Spreadsheet
Spreadsheet-style bulk research tool for running AI analysis across multiple companies in parallel.
Spreadsheet
Requires a Pro subscription.
Updated
pro
Overview
The Spreadsheet is a bulk research tool where companies or topics form rows and your research questions form columns. Fill individual cells, entire rows, or the whole sheet with AI-generated analysis. It supports templates, multi-sheet workbooks, document acquisition columns, and full export capabilities.
This is the power tool for analysts who need to compare the same data points across many companies at once. Build a template for your most common analysis pattern (e.g. "Earnings Quality Check") and reuse it every quarter. Results can be exported as CSV for Excel, HTML for formatted sharing, or PDF for archival.
How it works
The Grid
Subjects form rows, research questions and document requests form columns. Run cells individually or in bulk.
Subject | Risk Factors | Annual Reports |
|---|---|---|
AAPL | Supply chain concentration in... | 3 files |
MSFT | ◐ Researching (8s) | 3 files |
NVDA | ○ Click to research | ◐ Downloading... |
Two Column Types
Research Column
Column Title
Risk Factors
Query
What are the key risk factors for {company}?
AI researches each row subject and returns a text answer with sources
Document Acquisition Column
Document Type
Annual Report
Years Back
3
Format
PDF
Discovers and downloads actual files for each row subject
Document Acquisition Pipeline
When you run a document acquisition cell, the system goes through four stages to find and download the requested documents. This typically takes 1–3 minutes per cell.
1
Discover
AI finds investor relations pages and document URLs
→
2
Search
Web search fills gaps for missing years
→
3
Scrape
IR page is scraped for additional download links
→
4
Download
Each URL is downloaded and stored (up to 10 files)
Subject Types
Company
Autocomplete ticker lookup
AAPL
Topic
Free-form text
AI chip market
File
Upload a document
Q4_earnings.pdf
Cell States
Empty
○ Click to research
Queued
◔ Queued
Running
◐ Researching (12s)
Completed
Apple reported $94.8B in Q4...
Failed
✕ Research timed out
What you can do
Grid & Data
- Subject types: rows can be companies (with autocomplete ticker lookup), topics (free-form text), or uploaded files (PDF, DOCX, TXT, MD, CSV, HTML — up to 50 MB)
- Two column types: "Research" (AI query) and "Document Acquisition" (discovers and downloads actual documents)
- Run AI research per cell, per row, or in bulk across all empty cells — up to 5 cells run in parallel
- Cell states: empty, pending (queued), running (with elapsed timer), completed, or failed — each with a colour-coded visual indicator
- Cell detail panel: view full result with markdown rendering, source citations with excerpts, downloaded files with metadata, and copy / download buttons
Research Columns
- AI-powered research: each cell sends a query to an AI research agent using the row subject as context
- Query templates: choose from built-in templates (Company Overview, Going Concern Warning, Pending Litigation) or write a custom query
- {company} placeholder: use {company} in custom queries — it is automatically replaced with the row subject (company name, ticker, or topic) at execution time
- Sources and citations: completed cells include source URLs, excerpts, reasoning, and confidence scores — inspect them in the cell detail panel
Document Acquisition
- Multi-step AI pipeline: the system discovers, searches for, and downloads actual documents for each row subject — no manual hunting required
- How it works: the system searches for investor relations pages and direct document URLs, fills gaps for missing years via web search, scrapes the IR page for additional links, then downloads and stores each discovered file (up to 10 per cell)
- Configuration: specify a Document Type (free-text — e.g. "Annual Report", "10-K", "Proxy Statement"), Years Back (1–10, how many historical years to search), and Preferred Format (PDF, HTML, or Any)
- Timing: the pipeline runs asynchronously and typically takes 1–3 minutes per cell depending on the number of years and source availability
- Results: completed cells show a file count badge in the grid — open the cell detail panel to see individual files with size, year, format, and source URL, then download individually or as a ZIP archive
- Troubleshooting: if no documents are found, try broadening the Document Type (e.g. "Annual Report" instead of "10-K"), increasing Years Back, or setting Preferred Format to "Any"
File Subjects
- Upload documents as row subjects: switch the subject type selector to "File", then drag-and-drop or browse to upload a document
- Supported formats: PDF, DOCX, TXT, MD, CSV, HTML — maximum 50 MB per file
- How it works: the uploaded file is processed and its extracted text is used as context for research column queries, letting you run the same questions across a set of documents
Import
- Import from URL: paste any web page URL and AI extracts company names, tickers, exchanges, sectors, and countries — works well with stock screener results, fund holdings pages, and index constituent lists
- Import from file: upload a PDF, DOCX, CSV, TXT, or MD file containing company names
- Three extraction strategies: "AI Task" (recommended — fast AI-powered extraction), "Page Scraper" (full page scraping — better for JavaScript-heavy pages), "Comprehensive Search" (multi-source — slower but more thorough)
- AI instructions: optional guidance to focus extraction (e.g. "Swiss-listed companies only" or "Extract fund holdings from this quarterly report")
- Preview and select: extracted companies are shown in a table with name, ticker, exchange, sector, and country — select or deselect individual companies before importing
Templates
- Save as template: capture your current column structure (types, queries, document specs) as a reusable template
- Include subjects: optionally save the current row subjects alongside columns in the template
- System templates: pre-built templates from the Bollwerk team, shown on the home page under "System Templates"
- User templates: your saved templates appear under "Your Templates" and can be reused across new spreadsheets
Export
- CSV: column headers and cell values — best for importing into Excel or other analysis tools
- HTML: formatted table with styling — best for sharing via email or embedding in reports
- PDF / Print: opens the browser print dialog for print-ready output or Save as PDF
- Download files: for document acquisition cells, download individual files, all files as a ZIP archive, or all files sequentially
Tips
- ▸Build a template for your most common analysis (e.g. "Earnings Quality Check") and reuse it across new companies each quarter
- ▸Run a single cell first to verify your query produces the output you want, then bulk-run the rest of the row or sheet
- ▸The cell detail panel is your friend — use "Copy All" to grab content + sources in one go
- ▸Export as CSV for further analysis in Excel; use HTML export for formatted sharing
- ▸For document acquisition, use specific official terminology — "Annual Report" or "10-K" will produce better results than vague terms like "financial report"
- ▸Document acquisition runs asynchronously and can take several minutes per cell — start a batch and check back rather than waiting
- ▸The {company} placeholder in research queries works with any subject type — for topics it inserts the topic text, for files it inserts the filename
- ▸Use the File subject type to run the same research questions across a set of uploaded documents — useful for comparing contracts, prospectuses, or internal reports
- ▸Import from URL works well with stock screener results pages, fund holding lists, or any page that lists company names in a table