Working Smarter with AI: Skills, Second Brain and Agents

Building Your Personal Context and Second Brain

The quality of an AI answer depends less on a clever prompt and more on the context the AI has at its disposal. Context engineering means giving AI the right instructions, profile and knowledge up front, so it delivers consistent results without long prompts. You start with a personal profile of who you are and how you write and extend it into a Second Brain.

• From Prompt Engineering to Context Engineering
• Projects, Notebooks and Custom Instructions
• Creating an "About Me" Profile and a Writing Style Profile
• Reverse Prompting: Letting AI Ask the Questions
• Managing AI Memory
• Second Brain Principles: Capture, Organize, Distill, Express
• Structuring Notes for AI: Markdown, Metadata and Links
• Choosing a Tool: NotebookLM, Obsidian, OneNote and Loop
• LAB: Building Your Second Brain

Creating Skills

Skills let you describe once how a task should be done, such as writing a meeting report, preparing a quote or creating a presentation in your house style, after which the AI applies that procedure every time it is relevant.

• What is a Skill?
• Skills, Prompts, Agents and Plugins Compared
• Anatomy of a Skill: SKILL.md, Instructions, Examples and Resources
• Writing a Good Skill Description
• Testing and Improving Skills
• Sharing Skills within a Team
• Evaluating Third-Party Skills and Plugins
• LAB: Creating Your Own Skills

Connecting AI to Your Tools

AI becomes far more useful when it can work with your real data instead of text you copy and paste. Connectors and the Model Context Protocol (MCP) link AI assistants to mail, calendars, file storage, CRM systems and note-taking apps. MCP Apps go one step further and bring interactive forms and dashboards right into the conversation.

• Connectors and the Model Context Protocol (MCP)
• Connecting Mail, Calendar and Files
• Connecting Your Second Brain
• MCP Apps: Interactive Interfaces in the Chat
• Permissions and Access Control
• When to Use a Connector and When to Use a Skill
• LAB: Connecting AI to Your Tools

Delegating Work to AI Coworkers

AI coworkers such as Microsoft 365 Copilot Cowork, ChatGPT and Claude Cowork no longer just help you write: you describe the result you want, and the agent plans the work, carries out multiple steps and delivers finished documents, spreadsheets and presentations. Delegating well requires clear assignments, checkpoints and a critical review of the result.

• From Assistant to Coworker: Agentic AI
• Microsoft 365 Copilot Cowork, ChatGPT and Claude Cowork
• Describing Outcomes instead of Steps
• Plans, Approval Checkpoints and Reviewing Results
• LAB: Delegating a Multi-Step Task

Building Custom Agents

Custom agents combine instructions, knowledge, skills and tools into an AI assistant with a fixed role. The newest generation of agents, like ChatGPT workspace agents and Grok Bot, runs on its own cloud computer, keeps working when you are offline and can pick up requests from chat apps such as Teams or Slack.

• Instructions, Knowledge, Skills and Tools
• Always-On Agents: ChatGPT Workspace Agents and Grok Bot
• Scheduled Tasks and Recurring Routines
• Sharing and Governing Agents
• LAB: Building Your Own Agent