[ AI/Claude ] Anthropic에서 제공하는 수업듣기-1(Feat. Claude 101)

 

Wahat is Claude?

Understanding Claude's capabilities

Claude can help with a wide range of tasks that go far beyond simple question-and-answer interactions to assistant-like partnership that can both automate and augment your work. Here's a few things Claude excels at:

• Writing and content creation: Claude can collaborate with you on social media posts, professional emails, and complex reports. Because Claude is trained to take direction on personality and tone, you can iterate together on structure and clarity until your voice comes through clearly.
• Research and analysis: Claude helps you explore research angles, compile findings, and analyze data to surface meaningful insights. You can upload documents and Claude will help you make sense of complex information—this is enabled by Claude's large context window, which can ingest 200K+ tokens (about 500 pages of text or more), allowing Claude to consider extensive materials in a single conversation.
• Coding assistance: Claude Opus 4.5 is our most powerful model yet and the best coding model in the world. This strong performance on real-world coding tasks means Claude can help you write, debug, and explain code across multiple programming languages.
• Problem-solving and reasoning: Claude handles complex cognitive tasks, mathematical problems, strategic analysis, and research. Claude Opus 4.5 and Sonnet 4.5 are hybrid models offering two modes: near-instant responses and extended thinking for deeper reasoning. Extended thinking allows Claude to work through problems step-by-step, making it well-suited for tasks that require careful analysis.
• Learning new things: Whether you're learning a new skill, exploring unfamiliar domains, or working through complex challenges, Claude can adapt to your learning style and pace. Learning mode is a new Claude experience that guides your reasoning process rather than providing answers, helping develop critical thinking skills.

 

글쓰기랑 조사 및 분석에 자신이 있나보네요.
글쓰기에서는 500 페이지가 넘는 페이지를 분석할 수 있나봅니다.

문제-해결에서는 어떤 단계적 처리를 하여 문제를 푸나봐요.
추론의 고전이죠. stepy-by-step

 

 

Your first conversation with Claude

프롬프트

But you may ask, what is a good prompt? Before your next conversation with Claude, consider a few things:

1. Setting the stage: What is your role and what are your objectives? Is there context about your work that Claude should know about?
2. Defining the task: What action do you want Claude to take? Do you want Claude to write, analyze, build, or something else?
3. Specifying rules: What's the style or tone you want Claude to use? Are there examples that you can attach to show Claude what you're looking for?

1. 현재 너의 상태나 상황을 객관적으로 설명해줌으로서 claude가 상황을 파악할 수 있도록 할 것.

2. 그래서 그 상황에서 뭘 얻고 싶은지 정의 내릴 것.

3. 출력해내는 결과물이 어떤 톤엔 매너를 갖게 해주면 좋을지. 또는 예시를 첨부함으로서 유사하게 따라할 수 있도록.few shot prompting 기술

 

 

 

Opus 모델의 특징:
1. 가장 큰 사이즈의 모델
2. 하이브리드 추론 능력

Sonnet 모델의 특징:
1. 토큰 비용과 능력의 최적화 모델
2. 결과가 더 빠르게 얻을 수 있음

 

추론 확장

 

 

Iterating on Claude's responses

Conversations with Claude are meant to be iterative. Chaining bite-sized prompts together allows for a natural dialogue where you guide the conversation based on Claude's replies.

If Claude's first response isn't quite what you wanted, you have several options:

Ask follow-up questions: Build on Claude's response by asking for more detail, a different angle, or clarification. For example: "Can you expand on the second point?" or "That's helpful, but can you make it more concise?"

Provide feedback: Tell Claude what you liked and didn't like about its response. "This is good, but the tone is too formal. Can you make it more conversational?"

Redirect or restart: If Claude went in a different direction than you intended, simply steer it back. "Actually, I was asking about X, not Y. Let me clarify...". Worst case, restart your conversation in a new chat to fully refresh the context.

 

출력된 결과물이 만족스럽지 못했을때,

1. 특정 부분을 직접 언급해서 물어보세요 또는 더 명확한/구체적인 요구 사항을 말해주세요.

2. 결과물 중 좋았던 부분과 좋지 않았던 부분을 말해주세요

3. 구체적인 요구사항이나 원하는 방향성을 직접 작성해보세요. 아니면 다시 한번 재실행 보세요.