Claude AI Tutorial in Tagalog | Beginner's Guide

Summary of the Tutorial (Claude AI – Beginner Guide in Tagalog)

Who this tutorial is for

• Virtual assistants
• Developers/coders
• Entrepreneurs
• Investors
• Content creators
• Students
• Casual users who mainly use basic chat but may not know Claude’s other capabilities

---

1) Interface Walkthrough (Browser + Desktop)

• The tutorial first covers Claude’s web (browser-based) chat experience.
• It also explains Claude has a downloadable desktop version.
• The focus is on learning all interface tools/buttons, not just how to chat.

---

2) Incognito Chat (Privacy)

• Incognito mode prevents Claude from remembering the conversation.
• Regular chat stores memory, letting Claude recall context from earlier chats.

---

3) Chat Box Features: Attaching Files & Context

Claude prompts can include attachments using the plus icon, such as:

• Files/photos (PDF, Docs, Excel, Word, etc.)
• Screenshots
• Projects
• GitHub repositories (useful for developers to analyze codebases)

The tutorial highlights Claude’s broad scope for analyzing many types of inputs.

---

4) Skills (Custom Behaviors/Instructions)

What skills are

• Skills are reusable instruction sets that define how Claude should behave and respond.
• Skills can be created, saved, and reused to keep output consistent.

Customization & management

• Skills can be browsed/added.
• Manage Skills provides customization options (skills/connectors/plugins).
• A built-in skill creator exists, and users can create their own skills too.

Skill creation workflow (demo)

The presenter creates a skill aimed at content creators:

• Goal: Convert a script/caption into three strong “hook” variations (attention-grabbing intros).
• The skill creator asks clarifying questions, such as:
  • platform/context
  • hook style (e.g., avoiding “bold claims”)
  • whether to rewrite full captions or only generate hook variants
• Then the presenter runs a sanity check/audit using a sample caption (e.g., “daily posting for 3 years with little growth”), and Claude generates relevant hooks.
• Finally, the skill is packaged and saved for installation/use.

Reuse using “/” commands in chat

• After saving, the user can call the skill in chat using a slash command (e.g., “caption to hook”).
• The example shows providing a caption and receiving hook variations based on the saved instructions.

3 ways to create skills

1. Using Claude code/chat (skill creator): define behavior conversationally, then save.
2. Share via MD files (upload/download of skill instructions):
   • skills can be downloaded as documentation and uploaded by others.
3. Manual definition (“right skill instructions”):
   • The tutorial demonstrates naming rules and constraints, such as:
     • lowercase only, no spaces (spaces become dashes)
     • output format rules and word count constraints
   • Example skill: “real script writer”
     • Generates a 30-second short video script with structure:
       • 0–3s hook
       • problem
       • solution
       • call to action
       • 60–80 words total (~30 seconds)

---

5) Connectors (Connect Third-Party Apps)

What connectors do

• Connectors connect Claude to external tools/apps, such as:
  • Canva, Notion, Figma, Gmail
  • Google Drive/Calendar, Adobe, etc.
• Connectors require authorization (e.g., allow Claude to connect to Canva).
• Once connected, Claude can often use the tool without repeatedly saying “Canva” every time—by implying tasks that map to the connected tools.

Demo: Canva Design Generation

• Steps demonstrated:
  • Connect Canva
  • Ask Claude to create a 5-slide carousel with specified dimensions (example: 1920×1080)
  • Receive generated slide content/designs in Canva
• The tutorial also emphasizes:
  • privacy/security warnings (malicious prompts, data sharing concerns)
  • confirmation/permission flows for actions (e.g., allow/confirm repeat usage)

---

6) Plugins (Bundles: Skills + Connectors + More)

Concept

• Plugins are described as a bundle—like a department of capabilities.
• A plugin may include:
  • sub-agents
  • skills
  • connectors
  • MCP servers
  • CLI
  • hooks
  • slash commands
• The tutorial skips deep internal details (subagents/MCP/etc.) but explains the concept clearly.

Marketplace & installing plugins

• There’s a plugin marketplace on Claude’s site.
• Some plugins are Anthropic-verified; others may require caution.
• The presenter installs example plugins:
  • Marketing: campaign planning/content creation skills + connectors like Canva
  • Finance: finance-related skills + connectors like Gmail/Google Calendar

Creating plugins: 3 approaches

1. Add from a repository (e.g., GitHub) if you have one.
2. Add from Anthropic sources/marketplace entries.
3. Create with Claude:
   • Claude asks discovery questions, such as:
     • what the plugin should do
     • who it’s for (you/team/public)
     • content type (educational, tutorial, etc.)
     • biggest friction point
     • target topics
   • Claude then generates a plugin (example demo: “doddi content ideas” / “content ideas”).
   • In that specific demo build, the generated plugin included skills only (not MCP/agents/hooks/connectors).

Uploading downloaded plugin files

• Plugins are saved and used as files (example mentions an extension like .plugin).
• The presenter demonstrates drag-and-drop upload, then using the plugin in chat via its included skills.

---

7) Research vs Web Search

• Research: Claude provides deeper answers using available sources (slower).
• Web search: Claude pulls latest internet information for up-to-date responses.

---

8) Models, Tokens, and “Effort”

• The right panel includes model selection:
  • Haiku (fastest, lower intelligence)
  • Sonnet
  • Opus (slowest, most intelligent—examples like “Opus 4.8”)
• Tradeoff:
  • Higher intelligence model = slower responses
• Tokens:
  • Tokens increase with longer conversations, creating a practical “limit” feel.
• Effort:
  • Higher effort can improve answers even on lower-tier models, but may cost more speed/tokens.

---

9) Thinking Mode

• The Thinking toggle shows Claude’s reasoning/thought process during complex tasks.
• The presenter recommends:
  • most users don’t need it
  • use it when you want to learn how Claude arrived at the answer, not just the answer itself

---

10) Projects (Persistent Workspace/Memory)

What projects do

• Projects provide persistent context across multiple chats, reducing the need to repeat instructions.
• They’re useful for ongoing work such as:
  • client work
  • website/app writing projects
  • product or longer-term tasks

Demo: “Client Email Responder” project

Steps shown:

1. Create a project (name + description)
2. Add project instructions (tone, format constraints, sign-off)
3. Add project data (example text with properties and available tour days)
4. Respond to new inquiries with consistent memory

The demo shows Claude generating a professional email reply, including correct subject and closing/signature.

---

11) Artifacts (Interactive Interfaces)

What artifacts are

• Artifacts aren’t just previews—they create interactive UI elements Claude can generate.
• Demo: Claude generates a retro shooter arcade game as an interactive artifact.
• Artifacts can include:
  • gameplay features
  • controls
  • even sound/effects (as shown in the demo)

Artifact benefits

• Artifacts can be downloaded
• They can be published so others can access them

---

12) Next Planned Topics

• The presenter says Cloud Code / Cowork / Design will be covered in future uploads, since the browser tutorial is already long.

---

Main Speakers / Sources

• Main speaker/presenter: Doji (tutorial author/content creator; name appears repeatedly in narration and examples)
• Claude AI source: Anthropic’s Claude (referenced directly as “Claude” and as a product of Anthropic)
• Third-party example sources used in demos: Canva, GitHub, Gmail/Google services (accessed via connectors)