Protected Learning Content

AICPE Gurukul content is created for learning purposes. Printing, copying and unauthorised reuse are restricted.

AICPE Learning Hub 100 AI Tools for Every Need
Tool 03 of 100
Claude
Tool 03 of 100 | AI for Thoughtful Work

Learn Claude AI Practically

Collaborate with Claude to understand complex information, analyse documents, develop clear writing and build useful outputs—without surrendering evidence, privacy or human judgement.

Analyse DocumentsSummarise, compare and extract evidence
Improve WritingStructure, tone, clarity and refinement
Create ArtifactsBuild standalone content and tools
Think CriticallyCriteria, critique and verification
Learning Objectives

What You Will Be Able to Do

Use Claude for careful, structured work rather than accepting a polished first response.

Understand Claude

Explain Claude's role and distinguish chats, files, Projects and Artifacts.

Define Success

Set clear goals and evaluation criteria before refining a prompt.

Ground Analysis

Connect claims to supplied documents and identify missing evidence.

Work Responsibly

Protect sensitive content, verify important work and retain final accountability.

Foundation

What Is Claude?

Begin with a practical understanding of Anthropic's AI assistant.

1 An AI Collaborator for Complex Work

Claude is a conversational AI assistant created by Anthropic. It can help break down problems, analyse information, develop writing, work with data and code, and produce structured outputs through natural-language conversation.

Simple definition: Claude is an AI assistant designed to collaborate on thinking, analysis, writing, document work, coding and creation.

Claude is particularly useful when a task contains substantial context: a long brief, multiple documents, detailed instructions, evaluation criteria or several rounds of revision. The quality of the result still depends on the material and direction you provide.

Professional mindset: Do not ask only, “Can Claude write this?” Ask, “What evidence, criteria and review process will make the result fit for use?”
Official reference: Anthropic presents Claude as an AI for problem-solvers that can support complex challenges, data analysis, code and creation. See the official Claude overview.

2 Claude Is Not an Automatic Authority

Like other generative AI systems, Claude may misunderstand a request, overlook part of a file, make an incorrect inference or produce information that needs updating. Clear writing is not proof of factual accuracy.

Use for Drafting

Outlines, explanations, comparisons, edits and alternatives that a person will review.

Verify Important Claims

Current facts, laws, prices, calculations, citations and high-impact recommendations.

Do Not Delegate Accountability

Medical, legal, financial, safety, hiring and other consequential decisions require qualified human judgement.

Capability Map

Important Claude Workspaces and Features

Availability can vary by account, plan, device and organisational settings.

Conversations

Ask, clarify, critique and refine while maintaining the useful context of the current chat.

File Uploads

Work with supported documents and images for summaries, extraction, comparison and analysis.

Projects

Organise related chats with shared project knowledge and reusable project instructions.

Artifacts

Create substantial standalone content such as documents, code, diagrams, visualisations or interactive tools in a separate working area.

Web Search & Research

Where available, use current web information and deeper research workflows, then inspect the sources.

File Creation

Where enabled, Claude can create or edit spreadsheets, presentations, documents and PDFs through natural-language instructions.

Feature-awareness rule: Never promise a feature solely because it appears in a lesson. Confirm the current interface, account access, sharing rules and data policy before relying on it.
Official references: Anthropic's help material covers file uploads, Projects, Artifacts and file creation.
Core Skill

The O–C–M–F–E Prompt Formula

Claude prompting begins with a clear outcome and measurable success criteria.

3 Five Parts of a Professional Claude Prompt

ElementMeaningExample
O — OutcomeThe result you want to achieve“Create an executive summary and action plan.”
C — ContextAudience, purpose and constraints“For an institute director reviewing learner feedback.”
M — MaterialDocuments, data or examples to use“Use only the attached survey and policy note for factual claims.”
F — FormatStructure, tone and length“Use a one-page brief followed by a priority table.”
E — EvaluationSuccess criteria and checks“Every recommendation must connect to evidence; flag gaps and uncertainty.”
Create an executive summary and 30-day action plan for an institute director using the attached learner-feedback survey and policy note. Use only those files for factual claims. Begin with five findings, then provide a table containing evidence, proposed action, owner, priority and measurable indicator. Keep the tone constructive. Mark missing information as “not established” and list all assumptions separately.
Anthropic's prompting principle: Define success criteria and decide how you will test them before spending time repeatedly modifying the prompt.
Official reference: See Anthropic's prompt engineering overview.

4 Give Examples and Separate Instructions

For important recurring tasks, show Claude an example of an acceptable output. When a prompt contains source material, separate the instruction from the material using clear headings or tags.

<task>Rewrite the message for institute directors.</task> <requirements>Warm professional English; retain every confirmed date and amount; maximum 130 words; end with one clear action.</requirements> <source_message>[Paste the non-sensitive source text here]</source_message> <quality_check>List any ambiguous facts before producing the final version.</quality_check>
Why this helps: Clear boundaries reduce the chance that source text, examples and instructions are confused with one another.
Document Intelligence

Analyse Long Material Without Losing Traceability

Document work should create a path from claim back to source.

5 The Four-Pass Document Method

Pass 1: Inventory

Ask Claude to list the uploaded files, their apparent purpose, coverage and any unreadable or missing sections.

Pass 2: Extract

Request facts, dates, figures, obligations or themes with page, section, row or filename references where possible.

Pass 3: Analyse

Ask for comparisons, patterns, contradictions, risks and gaps—clearly labelled as analysis.

Pass 4: Produce

Create the final brief, plan or draft only after correcting extraction errors and agreeing on criteria.

Human Check

Inspect a sample of every important output category directly in the original material.

Evidence Record

Save the source list, verification notes and assumptions alongside the final deliverable.

Long-context caution: The ability to accept a large file does not guarantee that every detail is interpreted correctly. Ask Claude to identify coverage limits and test critical sections.
Practical Applications

Where Claude Can Create Value

Adapt these applications to your learning, office, institute or business setting.

Learning & Teaching

  • Explain difficult concepts
  • Create case-based practice
  • Review a learner's reasoning
  • Build structured study material

Document Analysis

  • Summarise and compare files
  • Extract requirements and actions
  • Find contradictions or gaps
  • Create evidence-led briefs

Writing & Editing

  • Develop professional drafts
  • Maintain voice while improving clarity
  • Critique logic and organisation
  • Create versions for different audiences

Business Thinking

  • Analyse permitted feedback
  • Compare strategies using criteria
  • Create operational checklists
  • Develop risk and action registers

Technical Work

  • Explain or review code
  • Plan technical projects
  • Debug with sufficient context
  • Create prototypes and artifacts

Freelancing

  • Improve research-assisted deliverables
  • Create client questionnaires
  • Develop proposals and briefs
  • Build repeatable quality workflows
Practical Experiments

Learn Claude by Doing

Use public or self-created, non-sensitive material.

Experiment 1: Evidence-Led Document Comparison

Step 1: Supply

Upload two public documents on the same topic and explain the comparison goal.

Step 2: Compare

Request a table of agreements, differences, missing points and source locations. Prohibit unsupported guessing.

Step 3: Audit

Manually verify six claims. Correct any mistake and ask Claude to rebuild the table.

Learning output: A verified comparison matrix with explicit evidence and uncertainty.

Experiment 2: Build and Refine an Artifact

Step 1: Define

Choose a useful standalone output: checklist, calculator concept, mini dashboard, diagram or interactive study aid.

Step 2: Create

Describe the users, controls, content, style and success criteria. Ask Claude to create an Artifact where available.

Step 3: Test

Use five realistic test cases, record problems and request a revised version. Review carefully before sharing.

Learning output: An improved artifact plus a test record showing how human feedback shaped the result.
Responsible Use

Protect Information and Preserve Accountability

Careful document work begins before a file is uploaded.

Privacy Checklist

  • Remove passwords, IDs and unnecessary personal details.
  • Confirm that you are authorised to upload every file.
  • Use organisation-approved accounts and retention settings.
  • Review sharing permissions for Projects and Artifacts.
  • Do not publish an Artifact containing private or licensed material.

Accuracy Checklist

  • Define evaluation criteria before drafting.
  • Connect important claims to source evidence.
  • Check current facts against authoritative sources.
  • Test calculations, code and interactive outputs.
  • Request uncertainty and missing information explicitly.
Sharing caution: A polished Artifact can be easy to distribute. Confirm audience, access, data, copyright, functionality and accuracy before publishing or sharing.
Real-Time Practical Assignment

Create a Source-Grounded Decision Brief

Demonstrate prompting, document analysis, critique and human verification in one workflow.

Assignment: Evidence to Action

Step 1: Define

Select a real but non-sensitive question and write five success criteria for the final brief.

Step 2: Analyse

Provide two or more permitted sources and use the four-pass document method.

Step 3: Deliver

Create a one-page brief, evidence table, action plan, assumptions list and human verification note.

Submission evidence: Goal, criteria, source inventory, original prompt, extraction, two refinements, claim verification, final brief and five-line reflection.

Success Criteria Worksheet

  1. Intended decision: ______
  2. Audience: ______
  3. Authorised sources: ______
  4. Five quality criteria: ______
  5. Facts requiring verification: ______

Self-Review Score

  1. Evidence traceability: ___ / 5
  2. Accuracy and coverage: ___ / 5
  3. Clarity and usefulness: ___ / 5
  4. Human improvement: ___ / 5
  5. Privacy and responsible use: ___ / 5
Common Mistakes

Mistakes Learners Should Avoid

Replace attractive shortcuts with reliable working habits.

Wrong Habits

  • Starting without success criteria
  • Uploading files without checking permission
  • Asking for a summary without source references
  • Assuming the whole document was understood perfectly
  • Publishing an untested Artifact
  • Trusting polished language without verification

Professional Habits

  • Define outcome and evaluation first
  • Inventory and sanitise source material
  • Separate extraction from interpretation
  • Use examples and structured instructions
  • Test outputs against realistic cases
  • Retain final human review and accountability
Knowledge Check

Quick Quiz: Claude AI

Answer all ten questions and submit.

1. What is the best professional way to use Claude?

Claude can assist thinking and creation, while the user retains judgement and responsibility.

2. In O–C–M–F–E, what does M represent?

Material includes the documents, data, examples or content Claude should use.

3. What should come before repeated prompt improvement?

Prompt refinement is meaningful only when the desired result is defined and testable.

4. What is a Claude Artifact?

Artifacts support substantial outputs such as documents, code, visualisations and interactive tools.

5. What is useful about a Project?

Projects can provide reusable knowledge and instructions across related chats.

6. What is the first pass of the document method?

Inventory establishes what material exists, its purpose and possible gaps.

7. Why should extraction and interpretation be separated?

Separation makes claims traceable and assumptions visible.

8. Which file is suitable for practice?

Only authorised, safe material should be uploaded.

9. What should happen before sharing an Artifact?

Shareable creations require both functional testing and responsible content review.

10. Who approves the final work product?

Human approval and accountability remain essential.
Quick Revision

Remember These Six Points

Review before moving to Tool 04.

1. Define Success

Set the outcome and evaluation criteria before refining prompts.

2. Supply Clean Context

Provide relevant, authorised material and clear source boundaries.

3. Structure Instructions

Separate the task, requirements, source material and quality check.

4. Analyse in Passes

Inventory, extract, analyse and produce rather than jumping to conclusions.

5. Test Artifacts

Check function, accuracy, access and rights before sharing.

6. Verify and Approve

Use source evidence and human judgement for the final result.

AICPE Quality Learning Commitment: AICPE Gurukul provides practical, skill-based and career-oriented learning for students, institutes and professionals. Learn more at aicpeindia.org and aicpe.online.