Enter passcode to access materials
Contact the instructor for access
Build your AI operating system — model, skills, tools, jobs
Everything from tonight lives here. Watch the replay, grab the diagnostic template, and keep this page open while you build.
Session Recording
Full Claude Masterclass walkthrough — posted after class
Claude Code — install before class
Our workspace for the whole session. Have it running before Friday.
We start where you are, then build the four components one layer at a time. By the end you'll have a working picture of your own AI operating system.
Not because it's hard — because nobody showed them the other 95%. You can't improve what you haven't named. So we score where you are on each of the four components, 1 to 5.
Do you pick the right model and brief it well, or take whatever's default?
Do you re-type instructions each time, or have reusable playbooks?
Is Claude connected to your real apps, or copy-pasting in and out?
Does anything run on autopilot, or do you start every task by hand?
Let Claude assess you. Paste this prompt, answer honestly, and it scores you 1–5 on each component and tells you the one move that lifts you a level.
Everything you build with Claude is some mix of these four. Together they turn a chatbot into an operating system that works for you.
The reasoning engine you talk to. Pick the right one for the job and brief it well.
Reusable playbooks you teach Claude once so it does the task your way every time.
How Claude reaches your real apps — Gmail, Calendar, Drive, the web — to read and act.
Cron jobs and automated triggers that run skills + tools on a schedule, without you asking.
The model is the reasoning engine. Two things matter: picking the right one for the task, and briefing it like a sharp new hire — role, context, and what "done" looks like.
Quick, cheap, instant. Great for simple lookups and high-volume tasks.
The everyday workhorse — smart and quick for most real work.
The heavyweight. Reach for it on hard reasoning and high-stakes thinking.
"You're my chief of staff." Setting the role sets the standard.
Paste the thread, the doc, the numbers. Don't make it guess.
Format, length, audience. "Three bullets for my board."
A skill is a playbook you teach Claude once — an SOP a specialist would follow — so it does the job your way every time, instead of you re-explaining it from scratch.
Every time you re-paste the same instructions, examples, and tone notes. Same setup, every time.
"Use my proposal playbook." Claude already knows the structure, the voice, and the standard.
A teammate who can't see your calendar, inbox, or files is guessing. Tools and connections let Claude read and act on your real apps — so it works from facts, not assumptions.
"Summarize everything unread this week and flag what needs a reply."
"What's my week look like? Find two hours of deep-work time and protect it."
"Pull the latest proposal and turn the pricing section into a clean table."
"Read this signup sheet and group people by role and seniority."
"Find my notes from the last three calls with this client and draft an update."
"Research this company and tell me who I'd actually want to meet."
A scheduled job runs a skill and its tools on a cron schedule (every weekday 8am) or an automated trigger (a new email from a key client). This is where Claude stops being something you use and becomes something that works for you.
Each morning: news, calendar, urgent email, and your top 3 priorities — one short brief.
Twice a day: sort mail into reply-now, FYI, ignore — with drafts ready for reply-now.
After every call: summary, decisions, and action items, sent to the right people.
The components aren't separate tricks — they stack. Here's a single morning brief that uses every one.
Sonnet does the reasoning — fast enough for a daily run.
Your "morning brief" format: 5 bullets, priorities first.
Reads your calendar and inbox to get the real facts.
Runs every weekday at 8am — before you're even up.
All four components get sharper when Claude has a memory of you — your notes, decisions, and context in one place. That's the second brain: a 4-week course building your personal AI assistant with Obsidian + Claude Code.
Obsidian
Where your notes live — plain files Claude can read
Open Claude Code. In the next 15 minutes, touch all four components. Use the timer in the corner.
The brain, its training, its hands, its heartbeat. Stack them and you've built an agent.
Take your weakest of the four and ship one scheduled job — the daily morning brief.
A 4-week course building your AI assistant with Obsidian + Claude Code.
Share your feedback
Two minutes — it shapes the next class
Stay in the loop
DayDreamers newsletter — new classes and drops
Have any questions? Just ask the instructor in the chat.