OpenClaw

Enter passcode to access materials

Contact the instructor for access

OpenClaw Workshop 1 / 15

Skills & Systems

Session 4 — From One-Off Prompts to Your AI Chief of Staff

Skills After-Meeting Follow-Up Daily Digest Memory Automation
press arrow to begin
Block 1: Recap 2 / 15
01 — Quick Recap

Session 3 Recap

Google Channels

Connected Gmail, Calendar, Drive, and Brave Search to your OpenClaw agent. One-time OAuth setup, full local control.

Real Workflows

Meeting notes pipeline, Digital Twin deal analysis, and LinkedIn post generation — all running end-to-end through your agent.

Today: We move from one-off prompts to repeatable, automated skills.
Block 1: The Vision 3 / 15
02 — The Vision

The AI Chief of Staff

I didn't get the best assistant I've ever had by asking better questions. I got it by giving the system a better operating model.

— Ryan Sarver

Memory Layer

Daily notes + MEMORY.md that compounds over time. The agent remembers everything.

Meeting Prep + Follow-Through

Briefs before meetings, action items after. Everything feeds back into memory.

Daily Rhythm

Morning brief, evening wrap. The system structures your day so nothing falls through the cracks.

1M+ views, 8,800+ bookmarks. This is where personal AI is heading.
Block 1: Where I Am 4 / 15
03 — Where I Am

Building in the Open

I've been experimenting with this stuff a lot. Things are constantly fluid and in motion.

Claude Code

What I use day-to-day for coding, automation, and skills.

OpenClaw

What we're learning together in this class — same core patterns.

Codex

OpenAI's version — the tools are converging on the same idea.

The tool doesn't matter as much as the pattern: persistent memory + real tool connections + repeatable workflows = your AI chief of staff.
Block 1: Key Concept 5 / 15
04 — Key Concept

What Are Skills?

A skill is a reusable workflow your agent can run on demand or automatically.

Example: SKILL.md File Structure

# After-Meeting Follow-Up ## Trigger Run after any meeting ## Context - Meeting transcript - Project info & team contacts - Past meeting history ## Steps 1. Parse transcript for decisions & action items 2. Draft personalized follow-up emails 3. Create tasks in tracker 4. Update daily note ## Output - Meeting recap (Google Doc) - Draft emails (Gmail) - Tasks (Todoist / Google Tasks)

Copy-Paste vs Running a Skill

ChatGPT (Copy-Paste)Running a Skill
ContextZero — you re-explain everythingAlready has your meetings, projects, contacts
EffortManual every timeRun once, works forever
OutputRaw text you copy-pasteDirectly creates emails, docs, tasks
MemoryForgets between sessionsCompounds over time
Block 2: Skill 1 — After-Meeting Follow-Up 6 / 15
05 — Skill 1

After-Meeting Follow-Up

Meeting Transcript
Zoom / notes
Parse Decisions & Actions
Skill engine
Draft Emails
Personalized
Create Tasks
With deadlines
Update Note
Daily note

What It Reads

  • Meeting transcript
  • Notion / project page
  • Attendee info & roles
  • Past meeting history

What It Outputs

  • Clean recap with decisions
  • Personalized emails per attendee
  • Tasks with deadlines
  • Daily note entry
The email to Jeffrey will be different from the email to Jason — the skill knows their roles.
Block 2: Live Demo — Building the Skill 7 / 15
06 — Live Demo

Building the Skill

  1. Open the SKILL.md file
  2. Walk through the prompt template
  3. Run it on the sample transcript
  4. Review the outputs: recap, emails, tasks
Follow along on your own machine!
Breakout: After-Meeting Skill 8 / 15
Breakout Room

Your Turn

Run the after-meeting skill on your own transcript

10 minutes
Block 3: Skill 2 — Daily Digest 9 / 15
07 — Skill 2

Daily Digest

Every morning you check 5+ apps. Every evening you wonder what you accomplished.

Calendar

Today's schedule, upcoming deadlines, meetings to prep for.

Email

Unread messages, flagged items, threads needing your response.

Meeting Notes

Action items assigned to you, open follow-ups from past meetings.

Calendar
+ Email + Notes
Daily Digest Skill
Process & prioritize
Morning Brief
/ Evening Wrap
Block 3: Live Demo — Daily Digest 10 / 15
08 — Live Demo

Running the Digest

  1. Show the skill configuration
  2. Run on the demo account
  3. Walk through the output
  4. Compare: morning brief vs evening recap
Notice how it catches things you'd miss — like that partnership email sitting unread for 6 days.
Breakout: Daily Digest 11 / 15
Breakout Room

Your Turn

Configure and run the daily digest on your own account

10 minutes
Block 4: Putting It Together 12 / 15
09 — Putting It Together

The Compound Effect

Each skill run adds to what your agent knows. Over time, it genuinely understands your work.

Day 1

Agent processes your first meeting. Learns names, roles, project context.

Week 1

5+ meetings processed. Patterns emerging — who owns what, recurring themes.

Month 1

Full context on every project, person, and commitment. Proactive suggestions.

The Chief of Staff

Not from one big moment, but from compounding context.

Context is the moat.
Block 4: Key Concepts 13 / 15
10 — Key Concepts

Today's Concepts

ConceptWhat It Means
SkillsRepeatable workflows with persistent instructions
Context AccumulationEach skill run adds to what your agent knows
Trigger-Based ExecutionSkills that run after meetings, daily, on new email
Personalized OutputSame skill, different output based on YOUR context
Human-in-the-LoopReview before send — especially emails and posts
Compound IntelligenceThe agent gets better the more you use it
Block 4: Homework 14 / 15
11 — Homework

This Week's Challenge

  1. Run the after-meeting skill on your next real meeting
  2. Set up the daily digest to run every evening for 5 days
  3. After 5 days: review your digests — what patterns emerge?
  4. Bonus: customize a skill for YOUR specific workflow
  5. Share your best skill output in the WhatsApp group
The goal: by next session, your agent should know your week better than you do.
Block 4: Next Time 15 / 15
12 — Next Time

Session 5 Preview

Capstone Project

Build your own custom skill from scratch. Design the trigger, context, steps, and output for YOUR workflow.

Voice Agents

Talk to your agent instead of typing. Real-time voice interaction changes everything.

Plus: sharing skills with others, and the OpenClaw ecosystem.

The best AI agents aren't the smartest — they're the ones embedded in your actual workflow.
05:00
Breakout Timer
05:00

Have any questions? Drop them in the Zoom chat.