Core Build: Process 20 Papers in 50 Minutes

Organize Documents by Source Type

Before uploading, categorize papers into three tiers for prioritized extraction:

Research Topic/
├── Primary Sources (10 papers)
├── Secondary Sources (7 papers)
└── Background (3 papers)

Primary Sources: Core research papers directly addressing the research question

Secondary Sources: Related studies providing context or comparative analysis

Background: Foundational texts establishing theoretical framework

Upload Documents to Claude Project

Open the Claude Project created in Quick Start.

Click Add Content in the project knowledge panel.

Verify Document Upload Quality

Test document accessibility before extraction:

Ask Claude: "List all documents in this project with author and publication year."

Expected response format:

1. Smith, J. (2023) - AI Research Methodology
2. Chen, L. (2022) - Quantitative Analysis Framework
3. [Continue for all 20 papers...]

Verify all 20 documents appear. If any are missing, re-upload and refresh project.

Create Document Index

Request a structured index from Claude:

Prompt: "Create a numbered index of all documents organized by: (1) Primary Sources (2) Secondary Sources (3) Background. Include author, year, and one-sentence description of each paper's core contribution."

Save this index as a text file or note. This becomes the reference for systematic extraction in Part 3.

Configure Project-Level Custom Instructions

Click Project Settings (gear icon) in Claude interface.

Scroll to Custom Instructions section.

Paste the following advanced instruction template:

Extraction Focus: Research questions, methodology, findings, limitations
Cross-Reference: Check for contradictions and supporting evidence
Citation Format: APA 7th edition with page numbers
Output Structure: Thematic organization with evidence hierarchy

What Each Directive Does:

• Extraction Focus: Ensures consistent information types from each paper
• Cross-Reference: Identifies agreements/conflicts between papers
• Citation Format: Standardizes all references for final bibliography
• Output Structure: Organizes findings by theme, not by paper

Add Research-Specific Guidance

Below the base template, add domain-specific instructions:

For Literature Reviews: "Identify methodological trends across papers and note evolution of theoretical frameworks."

For Systematic Reviews: "Extract effect sizes, sample sizes, and statistical significance for all quantitative findings."

For Meta-Analysis Prep: "Create comparison tables for variables measured across studies (methodology, sample characteristics, outcomes)."

Save custom instructions. Claude will apply these to all conversations in this project.

Test Custom Instructions

Create new conversation in the project.

Ask: "Summarize the first paper in the index according to project guidelines."

Verify Claude's response includes:

• Research question explicitly stated
• Methodology described with key details
• Findings organized thematically (not chronologically)
• APA citations with page numbers: (Author, Year, p. X)
• Limitations section identifying study constraints

If response lacks any element, refine custom instructions and retest.

Extract Information from Primary Sources (Papers 1-10)

Use systematic iteration prompt pattern:

For each paper in Primary Sources, extract:
(1) Core argument (2) Key evidence (3) Methodology
(4) Limitations (5) Citations to other works in corpus

Start with Paper 1. Paste prompt and review output.

Copy output to external document (Google Doc, Notion, or plain text file) labeled "Extraction_Primary.txt"

Repeat for Papers 2-10. Append each extraction to the same document.

Process Secondary Sources (Papers 11-17)

For secondary sources, adjust extraction focus:

Prompt: "For this secondary source, identify: (1) How it contextualizes primary research (2) Methodological comparisons to primary papers (3) Supporting or contradicting evidence."

Extract all 7 secondary sources using this modified prompt.

Save to separate file: "Extraction_Secondary.txt"

Extract Background Context (Papers 18-20)

Background papers require theoretical framework extraction:

Prompt: "For this background text, extract: (1) Key theoretical concepts (2) Definitions of core terms (3) Historical context for current research."

Process final 3 papers.

Save to: "Extraction_Background.txt"

Verify Extraction Completeness

Check all three extraction files contain:

• 10 primary source extractions (Extraction_Primary.txt)
• 7 secondary source extractions (Extraction_Secondary.txt)
• 3 background extractions (Extraction_Background.txt)
• Total word count: 6,000-8,000 words across all files

Ask Claude: "Count how many papers have been extracted and list any gaps."

If any papers are missing, return to that specific paper and complete extraction.

Upload Extraction Files to New Conversation

Create new conversation within the same Claude Project (keeps context fresh).

Upload all three extraction files:

• Extraction_Primary.txt
• Extraction_Secondary.txt
• Extraction_Background.txt

Verify files appear in conversation context (Claude confirms file access).

Identify Cross-Paper Themes

Prompt for thematic analysis:

"Analyze all extractions and identify 5-7 major themes that appear across multiple papers. For each theme, list which papers address it and whether they agree or contradict each other."

Expected output structure:

Theme 1: [Theme name]
- Papers addressing: 1, 4, 7, 11, 15
- Consensus points: [List agreements]
- Contradictions: [Note conflicts]
- Evidence strength: [Strong/Moderate/Weak]

Review theme list. Verify themes represent genuine patterns (not single-paper ideas).

Generate Thematic Synthesis Report

Request comprehensive synthesis organized by themes:

Prompt: "Write a 3,000-word research synthesis organized by the themes identified. For each theme: (1) Present consensus findings (2) Discuss contradictory evidence (3) Cite specific papers with page numbers (4) Note methodological limitations affecting conclusions."

Claude will generate synthesis document using all extraction data.

Review Synthesis Quality

Check synthesis document for:

• Clear thematic organization (sections for each theme)
• Evidence from multiple papers per theme (minimum 3 papers per theme)
• Balanced presentation of agreements and contradictions
• APA citations with page numbers throughout
• Transitions connecting themes to broader research question

Request revisions if any element is missing: "Expand Theme 3 with more evidence from secondary sources and add page number citations."

Generate Citation Verification List

Prompt: "Create a numbered list of every citation in the synthesis document, showing: (1) Citation text (2) Source paper (3) Page number (4) Theme it supports."

Expected output:

1. "AI improves research efficiency by 40%" (Smith, 2023, p. 15) - Theme: Productivity
2. "Methodology concerns persist" (Chen, 2022, p. 203) - Theme: Limitations

Save this list as "Citation_Verification.txt"

Spot-Check Random Citations

Select 5 random citations from verification list.

For each citation, ask Claude: "Show me the exact text from [Paper Name] page [X] that supports this claim."

Claude will quote source text. Verify the quote matches citation claim.

If discrepancy found, ask: "Correct this citation with accurate page number and quote."

Generate Bibliography

Prompt: "Create a complete APA 7th edition bibliography for all 20 papers cited in this synthesis, alphabetized by author last name."

Review bibliography for:

• Alphabetical order by first author surname
• Consistent APA formatting (italics, punctuation, capitalization)
• DOI or URL included where available
• All 20 papers present (count entries)

Final Citation Audit

Ask Claude: "Are there any claims in the synthesis document that lack citations? List them."

If Claude identifies uncited claims, either:

• Add citation from appropriate paper: "Add citation to this claim from Paper 7."
• Remove claim if no supporting evidence exists

Repeat until Claude confirms: "All claims are properly cited."