AI Document Type Classification automatically categorizes your uploaded internal documents
- What is AI Document Type Classification
- Financial Services Document Types
- Life Sciences & Healthcare Consulting Document Types
- What Happens During Content Ingestion
- How to Provide Feedback and Update Document Types
- Data Privacy and Security
What is AI Document Type Classification
AI Document Type Classification identifies the purpose of your documents and assigns them to predefined industry-specific categories.* This process improves search relevance and helps your team navigate internal content efficiently.
How It Works
When you upload a document, the system:
- Analyzes key sections of the document
- Determines the document’s primary purpose
- Assigns the most likely document type
The system uses a global classification framework that defines the relevant criteria for each document type. As your team provides feedback, it refines these prompts to create an account-specific taxonomy that continuously improves accuracy for your internal content. Your data remains private and is not used to train models for other customers.
Note: Available for Financial Services customers (as of March 27, 2026) and for Life Science & Healthcare Consulting Enterprise Intelligence customers and trialers (as of August 6, 2026).
Financial Services Document Types
The system classifies documents into the following categories:
| Document Type | Definition | Example |
|---|---|---|
|
Broker Research
|
Formal sell-side equity research issued by regulated brokerage firms, containing explicit ratings and price targets and ending with mandatory regulatory disclosures. | E.g. Research by Goldman Sachs, Morgan Stanley |
| Board Deck | A board deck is a presentation prepared for a company’s board of directors to provide updates on business performance and strategic direction. It typically includes financial results, key metrics, operational highlights, risks, and important decisions requiring board input. | |
| CIM | A sell-side transaction marketing document prepared by financial advisors to solicit buyer or investor interest, characterized by confidentiality language, investment highlights, and adjusted EBITDA with add-backs. | |
|
Company Documents
|
Primary-source documents authored by a company itself, including SEC filings, earnings releases, investor presentations, governance policies, and customer-facing operational materials. | E.g. Company Filing, Financial Reports, Annual Reports |
|
Company Tear Sheet
|
A concise, standardized snapshot of a company’s key facts—business description, financial metrics, valuation multiples, and ownership—designed for quick reference. | |
|
Contracts
|
Legal agreements governing business or transaction relationships between parties. These documents are formal, legal in tone, and focused on rights, obligations, and terms. |
E.g. NDAs, LOIs, purchase agreements
|
| Deal Process Management |
Operational documents used to manage and track the execution of a deal process. They are administrative in nature and help coordinate deal workflow rather than present the investment opportunity. | E.g. buyer lists, diligence checklists, process trackers, buyer reports |
| Earnings Summary |
Documents focused on a company’s earnings results for a specific period. This includes earnings call transcripts, analyst summaries, results breakdowns, and earnings presentations. These documents help analyze performance versus expectations and assess implications for the investment thesis. | |
|
Expert Calls
|
A transcript or summary of a conversation with an industry expert, consultant, or former executive, capturing qualitative insights rather than structured investment analysis. | E.g.expert call, interview, and key opinion leader transcript |
|
Independent Investment Research
|
Investment research produced by independent firms or platforms that are not sell-side banks or your internal team. These reports often include stock analysis, valuation perspectives, and investment theses. | E.g. Research from Value Investors Club or boutique research providers |
| Internal Investment Research |
Analysis created by your firm’s own team focused on evaluating specific companies or investment opportunities. These documents include proprietary insights, internal discussions, and evolving investment theses. They are used to support portfolio decisions and reflect the firm’s unique perspective. | |
| Internal Market Research | Internally curated or annotated content focused on industries, markets, or macroeconomic trends. These documents often combine external articles or data with internal commentary or summaries. Unlike investment research, they are not centered on a single company decision. | |
|
Investment Memo
|
An internal buy-side document designed to persuade investment decision-makers, featuring a structured investment thesis, valuation analysis (DCF/IRR), risks, and explicit internal recommendations. | |
|
Market Research
|
Market Research includes externally authored analysis of financial markets, industries, sectors, or specific companies. | E.g. Consulting or Third-Party Market Research Reports, Journalistic Market Analysis, Market Data Provider Outputs |
|
Meeting Notes
|
Notes captured from meetings, conferences, or discussions with management teams, investors, or industry participants. These documents summarize key points, observations, and insights from the interaction. They are generally informal and reflect real-time conversations rather than structured research. | |
|
Pitch Deck
|
A presentation used to communicate an investment opportunity, company story, or advisory capabilities. This includes investor presentations, management decks, and investment bank pitch books. Pitch decks are designed to persuade an audience and highlight key strengths, strategy, and growth potential. |
E.g. management presentation, roadshow deck, marketing deck, pitch
|
|
Teaser
|
A short, high-level document used to introduce an investment opportunity to potential buyers or investors. It typically provides a brief overview of the business, key highlights, and financial summary without revealing full details. Teasers are used early in deal processes to generate interest. | |
|
Tombstone
|
A brief announcement document highlighting a completed transaction. It typically includes details such as the companies involved, transaction type, and advisory roles. Tombstones are standardized and used for marketing or showcasing past deals. | |
|
Valuation Model
|
Financial models used to estimate the value of a company or investment. These are often spreadsheets containing projections, assumptions, and outputs such as DCF valuations or comparable analysis. They are analytical tools used to support investment decisions. | |
|
White Paper
|
In-depth research documents that explore a specific topic, theme, or methodology in detail. These are typically formal, structured, and analytical in nature. White papers are often used to present insights, frameworks, or thought leadership rather than specific investment recommendations. | |
|
Uncategorized
|
Documents that do not clearly fit into any defined category. This may include incomplete files, miscellaneous content, or documents with unclear purpose. These are typically edge cases or require further review to classify properly. | E.g. Web clippings, News articles, Informal notes, Spreadsheets, Images, Non-standard legal or administrative documents |
Life Sciences & Healthcare Consulting Document Types
The system also classifies documents for Life Science and Healthcare Consulting Enterprise Intelligence accounts into the following categories:
| Document Type | Definition | Example |
|---|---|---|
| Broker Research | Formal investment research reports published by banks or investment firms. | JPMorgan equity research report on a pharmaceutical company |
| Competitive Intelligence | The team's own strategic assessment of rival products and their positioning. | Spreadsheet comparing competitor drug features side-by-side with inclusion/exclusion reasoning |
| Conference Materials | Decks and materials built to run a live workshop, summit, or training session. | Annual summit agenda with speaker assignments and session descriptions |
| Contracts | Binding or draft legal agreements between parties. | Master Service Agreement (MSA) or Statement of Work (SOW) |
| Customer Calls | Interviews with end-users or decision-makers about their own experience and choices. | Interview transcript with a patient about their treatment experience |
| Disease Landscape | A characterization of a disease or therapy area, independent of any single product. | Overview document describing Type 2 diabetes prevalence, treatment options, and care gaps |
| Drug Asset Profile | A detailed profile of one or a few named products, covering mechanism, clinical data, and value story. | Slide deck walking through a drug's mechanism, trial results, and competitive advantages |
| Valuation Model | A financial spreadsheet with projections, assumptions, or historical data about a company or product. | Excel workbook projecting 5-year revenue for a new product |
| Earnings Summary | Coverage of a named company's earnings event, either analyst-written or the company's official announcement. | Company Q3 earnings press release or analyst write-up of results |
| Expert Calls | Interviews with outside experts consulted for their knowledge and expertise, not their personal purchasing decisions. | Call with a KOL discussing treatment patterns in their specialty |
| Market Research | A polished, finished research output—market sizing, segmentation, or synthesized findings from many respondents. | Report sizing the market for a therapeutic area across patient segments |
| Internal Market Research | The team's own working analysis of a market, sector, or macro trend—rough drafts and research inputs. | Analyst's personal notes on pricing-environment changes in the industry |
| Meeting Notes | A record of what a specific meeting discussed and decided. | Notes from a project status check-in with decisions and assignments |
| Project Engagement | A document that proposes or scopes a client project before it's contracted. | Consulting proposal outlining the approach and timeline for a market assessment |
| Regulatory | Official documents issued by a regulator or health authority. | FDA drug label or CMS coverage guidance document |
| Scientific Publication | Peer-reviewed research published in journals or presented at conferences. | Published clinical trial results in a peer-reviewed medical journal |
| Survey | A structured questionnaire or data-collection instrument given to respondents. | Questionnaire asking payers about their formulary and coverage processes |
| Go-to-Market | The commercial launch and market-access plan for a named product. | Product launch plan covering payer strategy, account targeting, and channel execution |
| Uncategorized | Documents that don't fit other categories—administrative files, templates, or raw data with no context. | Personal tracking sheet or empty project template |
What Happens During Content Ingestion
During content ingestion, the system analyzes a limited portion of each document to determine its type. This process prioritizes accuracy while protecting sensitive data.
What the System Uses for Classification
The system uses only the minimum information required to identify a document’s purpose:
- Selective document excerpts - the system reviews the first 2–5 pages to identify key signals.
- Document type definitions - predefined categories guide classification. These definitions improve over time based on your organization’s usage.
- High-level metadata - includes: Document title, Ingestion source, Company name
After analysis, the system assigns a document type. You can then:
- Confirm the classification
- Reject it
- Change it manually
If the system cannot determine the document type from the initial review, it:
- Selects additional relevant sections
- Analyzes those sections for stronger signals
- Repeats this process up to two times
What the System Does Not Use or Share
To protect sensitive information, the system does not use or expose:
- Full document content
- Internal document IDs
- User IDs or user-level identifiers
- Account-specific system references
How to Provide Feedback and Update Document Types
You can update a document’s classification directly in AlphaSense. This feature helps improve accuracy and ensures your organization’s taxonomy reflects your needs.
How to Update a Document Type
All users can update a document’s type from the Document Viewer:
- Open a document in Document Viewer
- Click the current Document Type
- Select Manage Document
- Choose the correct document type
- Save your changes
The system applies your update immediately.
Administrative Controls
Admins can manage document types at a broader level:
- Go to Settings → Preferences → Document Types
- Review existing classifications
- Take one of the following actions:
- Confirm classifications
- Reject classifications
- Change document types
- Request new document type categories
Your feedback and data remain private and secure:
- Nothing learned from your documents is shared with other accounts
- Every account maintains its own independently refined taxonomy
- No content from your account is used to train shared models
Data Privacy and Security
AlphaSense designs AI document classification to protect sensitive information. The system limits the data analyzed and ensures that your feedback remains private to your organization.
How the System Protects Your Data
The system applies the following safeguards during classification:
- Uses minimal data - only selected excerpts and high-level metadata are analyzed, never full documents
- Limits exposure of sensitive content - sensitive or irrelevant information are intentionally excluded from processing
- Stays within your account - classification uses your organization’s unique taxonomy and feedback .
What the System Does Not Use or Share
To maintain strict privacy standards, the system does not:
- Share your documents or insights with other customers
- Use your content to train shared or external AI models
- Expose internal identifiers, such as document IDs or user IDs
- Reference account-specific systems outside your environment