Skip to main content
Komo’s Structured Research transforms how professionals analyze large datasets and documents. Instead of researching entities one-by-one through chat interfaces, Structured Research lets you process hundreds—or even hundreds of thousands—of companies, documents, or data points simultaneously in a systematic table format with complete traceability and citations.

​
What is Structured Research?

Structured Research is a spreadsheet-like interface where AI becomes your research team, filling in information across rows and columns at massive scale. Unlike traditional chat-based AI that handles one question at a time, Structured Research enables systematic analysis of:
  • Thousands of entities (companies, people, products, markets)
  • Hundreds of thousands of documents (contracts, filings, reports, agreements)
  • Complex data extracted and synthesized with full citations
Two Core Research Types:
  • Enrichment Tables - Research entities and enrich with specific data points
  • Document Tables - Upload documents and extract answers systematically

​
Why Structured Research vs. Chat Interfaces

​
The Chat Interface Problem

Traditional AI chat (including ChatGPT) is designed for conversational, one-off queries: Limitations:
  • ❌ One question at a time = hours of repetitive prompting
  • ❌ No systematic organization of findings
  • ❌ Difficult to compare across multiple entities
  • ❌ Hard to verify sources for each claim
  • ❌ Results scattered across conversation history
  • ❌ Not scalable beyond ~10 items
Example Problem: 
"Research these 100 companies for funding status, founder backgrounds,
and competitive positioning"

ChatGPT: You'd need to ask about each company individually,
copy-paste 100 times, organize results manually,
lose citations, spend hours.

​
The Structured Research Solution

Komo’s table interface is built for systematic, scalable research: Advantages:
  • ✅ Research 100, 1000, or 1M entities simultaneously—all cells processed in parallel for maximum speed
  • ✅ Systematic organization in rows & columns
  • ✅ Side-by-side comparison built-in
  • ✅ Cell-level citations and explanations
  • ✅ Export to Excel/CSV for further analysis
  • ✅ 10-100x faster than sequential research thanks to parallel processing
Example Solution: 
Structured Research: Enter 100 company names in rows,
define columns (funding, founders, positioning),
click "Enrich" → complete in 15 minutes with full citations.

​
Enrichment Tables

​
What are Enrichment Tables?

Enrichment Tables let you research entities (companies, people, products, markets) by listing them in rows and defining research questions as columns. AI fills every cell with researched, cited information.

​
How It Works

1. Create Your Table Start at komo.ai/structured → New Enrichment Table 2. Define Rows (Entities) Enter entities to research:
  • Companies: “Cursor, GitHub Copilot, Codeium, Replit…”
  • People: “CEOs of Fortune 500 companies”
  • Products: “Top 50 SaaS project management tools”
  • Markets: “Emerging AI application categories”
Can enter:
  • Manually (type or paste)
  • Import from CSV/Excel
  • Generate from prompt (“Top 100 YC companies from W24 batch”)
3. Define Columns (Research Questions) Create columns for data points you want extracted:
  • “Tagline / One-liner”
  • “Key Differentiators / Moat Thesis”
  • “Target Segments (Indie/SMB/MM/Enterprise)”
  • “Lead Investor”
  • “Latest Funding Round”
  • “Headquarters Location”
4. Select Cells to Enrich Select the specific cells you want to enrich by:
  • Selecting individual cells
  • Selecting entire rows or columns
  • Selecting multiple ranges
  • Or selecting all cells in the table
5. Click “Enrich” AI researches the selected cells in parallel, filling them with:
  • Accurate, up-to-date information
  • Inline citations
  • Source links
6. Verify with Cell-Level Citations Click any cell to see:
  • Source Information Panel (right sidebar)
  • Exact value extracted
  • Confidence level (High/Medium/Low)
  • Reasoning explaining how AI arrived at answer
  • 4+ sources with direct links
  • Preview quotes from each source

​
Real-World Example: VC Deal Sourcing

Scenario: VC analyst evaluating AI coding startups Enrichment Table Setup: Rows: 29 AI coding companies (Cursor, GitHub Copilot, Codeium, Replit, etc.) Columns:
  • Company
  • Tagline / One-liner
  • Key Differentiators / Moat Thesis
  • Key Risks/Unknowns
  • Target Segments (Indie/SMB/MM/Enterprise)
  • Lead Investor
  • Latest Funding Amount
  • Founding Year
  • Team Size
  • GitHub Stars (if open source)
  • Pricing Model
  • Key Integrations
  • Customer Reviews Summary
  • Competitive Advantages
Process:
  1. Import list of 29 companies
  2. Define 14 research columns
  3. Select all cells to enrich
  4. Click “Enrich”
  5. AI researches 406 data points (29 × 14) in parallel
  6. Complete in 15 minutes
Cell-Level Verification: Click on “Codeium - Key Differentiators”: Source Information Panel: 
Value: "Offers multiple AI assistance modalities (Autocomplete,
        Chat, Command, Cascade); broad editor and language support;
        robust enterprise-grade features like self-hosting,
        zero-data-retention, and flexible deployment options."

✅ High Confidence

Reasoning:
Codeium's key differentiators include its multiple AI assistance
modalities (Autocomplete, Chat, Command, Cascade), broad support
for editors and languages, and robust enterprise-grade features
like self-hosting, zero-data-retention, and flexible deployment
options. Its own coding engine, Cortex, capable of processing
large codebases, may also contribute to efficiency.

Sources (4):
📄 Codeium: AI Coding Assistant for Autocomplete, Chat...
   skywork.ai

📄 Codeium AI: Navigating Enterprise Coding Challenges
   cisin.com

📄 This AI Coding Engine Can Process 100 Million Lines...
   forbes.com

📄 97% Cheaper, Faster, Better, Correct AI — with Varun...
   latent.space
Result:
  • Complete competitive landscape in single view
  • Every claim verifiable with sources
  • Export to pitch deck or investment memo
  • Share with team for collaborative review

​
Document Tables

​
What are Document Tables?

Document Tables let you upload dozens, hundreds, or thousands of documents, then extract structured information by asking questions across all documents simultaneously.

​
How It Works

1. Create Document Table Start at komo.ai/structured → New Document Table 2. Upload Documents Upload files in bulk:
  • Contracts: NDAs, MSAs, vendor agreements
  • Financial docs: 10-Ks, earnings calls, investor decks
  • Legal: Depositions, briefs, discovery documents
  • Research: Market reports, analyst notes, expert calls
  • Internal: Memos, presentations, strategy documents
Supported formats: PDF, Word, PowerPoint, Excel, images (OCR), text files Scale: Upload 1 to hundreds of thousands of documents 3. Define Research Columns Ask questions you want answered across all documents:
  • “What is the termination clause?”
  • “What are the liability limits?”
  • “Who are the key decision makers mentioned?”
  • “What are the main risks identified?”
  • “What is the projected market size?”
4. Select Cells to Enrich Select the cells you want to enrich (individual cells, rows, columns, or entire table). 5. Click “Enrich” AI reads every document and extracts answers for the selected cells in parallel. 6. Click Cells for Evidence Every cell shows:
  • Extracted answer
  • Direct quote from source document
  • Page number reference
  • Confidence level
  • Supporting context

​
Real-World Example: M&A Due Diligence

Scenario: Corporate development team reviewing acquisition target Document Table Setup: Documents Uploaded (87 files):
  • Financial statements (5 years)
  • Customer contracts (42 contracts)
  • Employee agreements (28 employees)
  • IP and patent documents (8 filings)
  • Board meeting minutes (4 years)
Research Columns:
  • Document Name
  • Document Type
  • Key Terms Summary
  • Red Flags Identified
  • Financial Obligations
  • Termination Clauses
  • Liability Caps
  • Revenue Recognition Terms (contracts)
  • IP Ownership Clarity
  • Material Risks Disclosed
Process:
  1. Upload 87 due diligence documents
  2. Define 10 extraction columns
  3. Select all cells to enrich
  4. Click “Enrich”
  5. AI analyzes 870 data points in parallel across all docs
  6. Complete in 30 minutes
Cell-Level Verification: Click on “CustomerContract_Acme.pdf - Red Flags”: Source Information Panel: 
Value: "Auto-renewal clause with 90-day cancellation notice;
        Liability capped at 1x annual fees (below industry standard);
        Customer owns all data and IP created during engagement;
        Contract includes audit rights with 30-day notice."

Sources (1):
📄 CustomerContract_Acme.pdf
   Page 7, Section 8.3: "Either party may terminate this agreement
   with 90 days written notice. Upon expiration of the initial term,
   this Agreement will automatically renew..."

   Page 12, Section 11.1: "Vendor's total liability shall not exceed
   the total fees paid by Customer in the twelve (12) months preceding..."

   Page 15, Section 13.2: "All work product, data, and intellectual
   property created or derived during the performance of Services shall..."
Result:
  • 87 documents systematically analyzed
  • Key risks surfaced automatically
  • Every finding backed by document excerpt
  • Legal team focuses on flagged issues only
  • Due diligence time reduced from weeks to days

​
Use Cases by Industry

​
Investment Management

Deal Sourcing
  • Research 500 companies in target sector
  • Columns: Funding, growth metrics, team backgrounds, competitive moat
  • Systematic pipeline building
IC Memo Creation
  • Upload 30 documents from VDR
  • Extract: financials, risks, opportunities, management commentary
  • Auto-generate first draft of investment memo
Portfolio Monitoring
  • Upload quarterly reports from 40 portfolio companies
  • Extract: KPIs, risks, guidance changes, management concerns
  • Quarterly board prep automated
Contract Review
  • Upload 200 vendor contracts
  • Extract: liability caps, termination terms, non-standard provisions
  • Identify outliers requiring negotiation
Litigation Support
  • Upload 500 pages of deposition transcripts
  • Extract: contradictions, key admissions, timeline of events
  • Build case strategy systematically
Regulatory Compliance
  • Upload 1000 customer contracts
  • Check: GDPR compliance, data retention terms, jurisdiction clauses
  • Proactive compliance audit

​
Corporate Development

M&A Screening
  • Research 100 potential acquisition targets
  • Columns: Revenue, growth, technology stack, customer overlap
  • Prioritize top 10 for deep dive
Vendor Analysis
  • Upload RFP responses from 15 vendors
  • Extract: pricing, capabilities, implementation time, references
  • Objective comparison matrix
Competitive Intelligence
  • Research 50 competitors
  • Columns: Product updates, funding, customer reviews, pricing changes
  • Quarterly competitive landscape report

​
Consulting

Market Research
  • Upload 40 industry reports
  • Extract: market size, growth drivers, key players, trends
  • Synthesize into client-ready insights
Expert Call Synthesis
  • Upload transcripts from 25 expert interviews
  • Extract: key themes, contradicting views, consensus opinions
  • Build point-of-view document
Benchmarking
  • Research 100 companies in client’s industry
  • Columns: Operational metrics, best practices, digital maturity
  • Identify improvement opportunities

​
Real Estate

Property Analysis
  • Research 200 properties in target market
  • Columns: Valuation, cap rate, occupancy, tenant quality, risks
  • Investment pipeline prioritization
Due Diligence
  • Upload 50 property documents per deal
  • Extract: lease terms, rent rolls, violations, maintenance issues
  • Comprehensive property profile
Market Tracking
  • Research 100 comparable sales
  • Columns: Sale price, price per sq ft, buyer, seller, terms
  • Market trend analysis

​
Key Features & Differentiators

​
1. Massive Scale with Parallel Processing

  • Komo: Process 1, 1000, or hundreds of thousands of entities/documents simultaneously in parallel
  • Chat interfaces: Practical limit of ~10 before becoming unusable, processed sequentially

​
2. Systematic Organization

  • Komo: Table format enables side-by-side comparison, sorting, filtering
  • Chat interfaces: Linear conversation, difficult to compare findings

​
3. Cell-Level Citations

  • Komo: Click any cell to see sources, reasoning, confidence
  • Chat interfaces: Citations (if any) at response level, not claim level

​
4. Verifiable Research

  • Komo: Every cell has confidence score + 4+ sources with direct quotes
  • Competitors (Hebbia): Source attribution but less granular verification

​
5. Export & Collaboration

  • Komo: Export to Excel/CSV for presentation or further analysis
  • Komo: Share tables with team, collaborate on research

​
6. Speed at Scale

100 entities × 10 columns = 1000 research points
  • Komo Structured Research: 15-20 minutes
  • Manual research: 50+ hours
  • Chat interface: 5-10 hours of repetitive prompting

​
Best Practices

​
Enrichment Tables

Be Specific with Column Names:
  • ✅ “Target customer segment (Indie/SMB/MM/Enterprise)”
  • ✅ “Latest funding round (Series, amount, lead investor, date)”
  • ❌ “Customers” (too vague)
  • ❌ “Funding” (ambiguous - total raised vs. latest round?)
Start with Core Columns, Expand Iteratively:
  • Create table with 3-4 essential columns
  • Enrich and review results
  • Add more columns based on findings
  • Enrich only new columns (faster)
Use Confidence Scores:
  • High Confidence (✅): Use directly
  • Medium Confidence (⚠️): Verify sources
  • Low Confidence (❌): May need manual research
Combine with Other Tools:
  • Enrich basic data in Structured Research
  • Use Deep Search for complex follow-ups
  • Export to Content Creation for presentations

​
Document Tables

Organize Documents by Type:
  • Create separate tables for contracts vs. financial docs vs. legal docs
  • Different document types = different relevant columns
Specific Extraction Questions:
  • ✅ “What is the total contract value in USD?”
  • ✅ “List all termination clauses with notice periods”
  • ❌ “Summarize the contract” (too broad)
Verify Key Findings:
  • Always click cells for high-stakes decisions
  • Review direct quotes from source documents
  • Check page numbers for context
Use for Pattern Recognition:
  • Sort/filter to identify outliers
  • “Which contracts have liability caps below $1M?”
  • “Which agreements have auto-renewal clauses?”

​
Comparison: Komo vs. Hebbia vs. ChatGPT

CapabilityKomo Structured ResearchHebbiaChatGPT Plus
Entity Enrichment✅ Up to 1M entities❌ Document-focused⚠️ Manual, <10 entities practical
Document Analysis✅ Unlimited documents✅ Unlimited documents⚠️ Limited to conversation context
Table Interface✅ Spreadsheet-like✅ Table format❌ Chat only
Cell-Level Citations✅ Click any cell⚠️ Document-level citations⚠️ Response-level citations
Confidence Scores✅ High/Medium/Low per cell❌❌
Export✅ Excel, CSV✅ Export available❌ Copy-paste only
Scale✅ Hundreds of thousands of entities/documents✅ Large document sets❌ Conversation-based
Speed (100 entities)15 minsN/A5+ hours
Reasoning Transparency✅ Per cell⚠️ Per document⚠️ Per response
When to Choose Komo:
  • Researching many entities (companies, people, products)
  • Need systematic comparison across entities
  • Want granular cell-level verification
  • Exporting to Excel for presentations/analysis
When Competitors Excel:
  • Hebbia: Complex legal document workflows, enterprise legal teams
  • ChatGPT: Conversational exploration, creative tasks, one-off questions

​
Common Questions

Q: How many entities can I research at once? A: Up to hundreds of thousands of entities. Practical limits depend on column complexity. Standard: 100-1000 entities works smoothly. All enrichments run in parallel for maximum speed. Q: How many documents can I upload? A: No hard limit. Users have analyzed 10,000+ documents in a single table. Upload time depends on file sizes. Q: How long does enrichment take? A: Depends on rows × columns. Generally: 100 entities × 10 columns = 15-20 minutes. 1000 entities × 10 columns = 2-3 hours. Q: Can I add more columns after initial enrichment? A: Yes. Add columns anytime and enrich only new columns (faster than re-enriching entire table). Q: What if a cell has wrong information? A: Click cell to see sources. If sources are weak, manually edit cell. Report issue to improve AI accuracy. Q: Can I filter and sort like Excel? A: Yes. Full spreadsheet functionality: sort, filter, search within table. Q: What file formats for documents? A: PDF, Word (.docx), PowerPoint (.pptx), Excel (.xlsx), images (with OCR), plain text. Q: How are documents processed? A: OCR for images/scanned PDFs, text extraction for digital docs, intelligent parsing of tables and structured content. Q: Can I share tables with my team? A: Yes. Share link or export. Collaborate by adding team members (permissions: view/edit). Q: Does this work for non-English documents? A: Yes. Supports major languages. Specify language if needed: “Extract in Spanish.” Q: How is this different from uploading docs to ChatGPT? A: ChatGPT processes documents conversationally. Komo extracts structured data across ALL documents systematically in table format with cell-level citations. Q: Can I use my own data sources (internal databases)? A: Yes. Integrate via API or upload exported data. Enterprise: custom integrations available.
Structured Research transforms how professionals handle large-scale research—enabling systematic analysis of millions of entities or documents with complete traceability, turning weeks of manual work into minutes of automated, verifiable research.