Real-World Use Cases: Research, Writing, and Coding

This lesson shows you how to set up Claude Projects for three common workflows: academic research, content writing, and software development. Each example includes the custom instructions, knowledge files, and conversation strategies that make the project effective.

Use Case 1: Research Project

Scenario

You are researching the impact of AI on hiring practices for a report. You have several academic papers, industry reports, and interview transcripts to analyze.

Project Setup

Project name: "AI Hiring Research"

Custom instructions:

Role: You are a research assistant helping me analyze
the impact of AI on hiring practices.

Context:
- This is for a 5,000-word industry report
- Target audience: HR directors at Fortune 500 companies
- I need data-driven insights, not opinions

Guidelines:
- Cite specific papers and page numbers when referencing uploaded documents
- Flag any conflicting findings between sources
- Distinguish between correlation and causation
- Note sample sizes and methodology limitations

Format:
- Use APA citation format
- Structure findings by theme, not by source
- Include a confidence level (high/medium/low) for each finding

Knowledge files to upload:

• 3-5 key academic papers (PDF)
• Industry survey data (CSV or PDF)
• Your report outline (Markdown)

How to Use This Project

Conversation 1 — Initial Analysis:

Read all the uploaded papers and create a comparison matrix:
columns for each paper, rows for methodology, sample size,
key findings, and limitations.

Conversation 2 — Theme Extraction:

Based on the papers, identify the top 5 recurring themes
about AI in hiring. For each theme, list which papers
support it and any contradictions.

Conversation 3 — Draft Writing:

Using the themes from our previous analysis and following
my report outline, draft the section on "Bias in AI
Screening Tools" with proper citations.

Use Case 2: Content Writing

Scenario

You manage a company blog and need to produce 3-4 articles per week that follow your brand voice and content strategy.

Project Setup

Project name: "Company Blog — 2026"

Custom instructions:

Role: You are a content writer for TechFlow, a project
management SaaS for remote engineering teams.

Audience:
- Engineering managers at 50-500 person companies
- Technical but time-constrained
- Skeptical of marketing fluff

Brand voice:
- Direct and practical — no filler phrases
- Use "you" instead of "users" or "one"
- Short paragraphs (2-3 sentences max)
- Include specific examples and numbers
- Occasional dry humor is fine; forced enthusiasm is not

Content rules:
- Every post needs a hook in the first sentence
- Include at least one data point or real example per section
- End with 3 actionable takeaways
- Target word count: 1000-1500 words
- SEO: Include the target keyword in the title, first paragraph,
  one H2, and naturally throughout

Never do:
- Use words like "leverage," "synergy," "game-changer"
- Start sentences with "In today's fast-paced world"
- Make unsubstantiated claims about productivity gains
- Mention competitors by name

Knowledge files to upload:

• Brand style guide (PDF or Markdown)
• Top 10 performing blog posts (for voice reference)
• Content calendar with upcoming topics
• SEO keyword research spreadsheet

How to Use This Project

Conversation 1 — Outline:

I need a blog post targeting the keyword "async standup meetings."
Draft an outline with a compelling hook, 4-5 sections, and
takeaways. Reference our style guide for tone.

Conversation 2 — Full Draft:

Write the full post from the outline we created. Make the
hook about the pain of 9 AM standups for distributed teams.
Include the async meeting template from our product as
a natural example (not a hard sell).

Conversation 3 — Edit Pass:

Review this draft for brand voice consistency. Flag any
sentences that sound too formal or marketing-heavy.
Suggest specific rewrites.

Use Case 3: Software Development

Scenario

You are building a full-stack web application and want Claude to understand your architecture, coding standards, and existing patterns.

Project Setup

Project name: "Dashboard App — Backend"

Custom instructions:

Role: You are a senior backend developer on my team.

Tech stack:
- Node.js with Express
- TypeScript in strict mode
- PostgreSQL with Prisma ORM
- Jest for testing
- Docker for deployment

Architecture:
- Clean architecture with controllers, services, and repositories
- All business logic lives in services, never in controllers
- Controllers handle HTTP concerns only (parsing, validation, response)
- Repositories abstract database queries

Coding standards:
- Every function has TypeScript types for parameters and return values
- All async functions use try/catch with specific error types
- No "any" types — use "unknown" and narrow
- Database queries use parameterized inputs (no string concatenation)
- Every new endpoint needs unit tests and integration tests

Response format:
- Show complete files, not snippets
- Explain architectural decisions with tradeoffs
- Suggest tests for any new code
- Flag potential security issues proactively

Knowledge files to upload:

• prisma/schema.prisma — Database schema
• src/types/index.ts — Shared TypeScript types
• Key service files showing existing patterns
• API documentation or OpenAPI spec
• .env.example — Environment variable structure

How to Use This Project

Conversation 1 — New Feature:

I need to add a user invitation system. Users should be able
to invite others by email, and invitations expire after 48 hours.
Design the database schema changes, API endpoints, and service layer.

Conversation 2 — Implementation:

Implement the invitation service based on the design from our
previous conversation. Follow the patterns in the existing
user service I uploaded.

Conversation 3 — Testing:

Write unit tests for the invitation service. Cover the happy path,
expired invitations, duplicate invitations, and invalid email formats.
Follow the testing patterns in the existing test files.

Building Your Own Project

Use this template to plan any project:

Planning Checklist

1. What is the project's purpose? (one sentence)
2. What role should Claude play? (researcher, writer, developer, analyst)
3. What context does Claude need? (background, constraints, preferences)
4. What files will Claude reference? (documents, code, data)
5. What are the rules? (formatting, tone, technical standards)
6. What should Claude avoid? (common mistakes, forbidden patterns)

Starting Template

Role: You are a [role] helping me [purpose].

Context:
- [Key background information]
- [Constraints or requirements]
- [Target audience or stakeholders]

Guidelines:
- [Rule 1]
- [Rule 2]
- [Rule 3]

Format:
- [How to structure responses]
- [Length or detail preferences]

Avoid:
- [Common mistake 1]
- [Common mistake 2]

Key Takeaways

• Research projects benefit from uploading source papers and asking Claude to compare, synthesize, and draft sections across separate conversations
• Writing projects need brand voice instructions, style guides, and top-performing examples as knowledge files
• Development projects work best when you upload your schema, key code files, and define architectural patterns in instructions
• Plan any project by defining Claude's role, context, guidelines, format, and constraints
• Use separate conversations within a project for different phases: planning, drafting, reviewing