When should you use it?
- You juggle several roles and lots of tools
- Morning prep eats time you need elsewhere
- You prefer a curated set of tasks to a long template list
Not for: Any task that sends messages without your review; Regulated data you may not route through these connectors
What do you need, and what do you get?
You need
- Connectors for your mail, calendar, chat and CRM
- Your roles and weekly rhythm
- Preferences for the dashboard layout
You get
- A short list of recommended scheduled tasks with reasons
- One morning page holding prepared drafts and to-dos
How does it work, step by step?
- Pick interview depthChoose a quick or thorough intake depending on time.
- Answer about roles and toolsCover what hats you wear, what is connected, your schedule and key people.
- Accept only fitting tasksAsk it to explain why each task suits you and what it skipped.
- Confirm before creatingReview the list before any schedule is set up.
- Keep drafts as draftsLet it prepare emails and messages, but send them yourself.
The prompt
The original prompt by Christian Pean MD, MS, word for word, with a link to where it was published. Below it is our adapted version with fill-in blanks, if you want it. Works in Claude Cowork and Claude. Values in lime are placeholders.
Original prompt by Christian Pean MD, MS
You are building me a personalized agentic operating system using Claude Cowork scheduled tasks. Conduct a structured interview, then produce a curated task portfolio matched to my actual roles and workflows — plus a personalized Jarvis Command Center dashboard. START by asking how deep I want to go: - Express (5 min): Role, tools, schedule, priorities. Get started fast. - Standard (15 min): Full professional context mapping. - Comprehensive (30-40 min): Full chief-of-staff intake. Then interview me across these domains (adapt to chosen depth): 1. IDENTITY & ROLES — What hats do I wear? (Clinician, executive, researcher, investor, parent?) 2. TOOLS & PLATFORMS — What's connected? (Gmail, Calendar, Slack, CRM, Granola, banking?) 3. SCHEDULE & RHYTHMS — What does my week look like? When do I need intelligence delivered? 4. STAKEHOLDERS — Who matters? (Board, investors, team, patients, partners?) 5. ROLE-SPECIFIC CONTEXT — Go deeper on whichever roles apply: clinical workflow, company building, finance/treasury, policy tracking, academic/learning 6. COMMAND CENTER PREFERENCES — What does my ideal morning dashboard look like? Dense or minimal? Dark or light mode? What do I want to see first? TASK CURATION (important): After the interview, do NOT recommend every available task. Instead: - Identify which personas apply: Essential, CEO, Clinician, Knowledge Worker, Healthcare Learner, Finance, Policy - Recommend ONLY the tasks that match my actual workflow and tools - For each, explain WHY it's relevant to me specifically - Call out what you're skipping and why - Always include the Jarvis Command Center — the daily dashboard that aggregates everything - Let me confirm or adjust before creating anything AVAILABLE TASKS (34 across 8 categories): ⭐ Command Center (1): Jarvis Command Center Essential (3): Email Briefing, Meeting Prep, Evening Debrief CEO/Executive (4): Deal Pipeline, Competitive Intel, Investor Update, Team Pulse Clinician (4): Clinical Prep, Literature Surveillance, CME Tracker, Quality Pulse Knowledge Worker (4): Priority Stack, Project Rollup, Stakeholder Update, Document Queue Healthcare Learner (5): Concept Card, Academic Crunch, Board Sprint, Journal Club, Teaching Prep Finance (4): Market Pulse, Financial News Digest, Earnings Tracker, Treasury Review Policy (4): CMS Radar, FDA Device Intel, Payer Monitor, Policy Briefing After I confirm my portfolio, generate: 1. second-brain.md — My structured context file with all interview answers 2. Curated task portfolio with personalization notes 3. Jarvis Command Center customized to my preferences 4. All scheduled tasks deployed and ready to run Present the recommendation as a table: | Task | Category | Why It's For You | Cadence | Then ask: "Here's your recommended portfolio of [N] tasks out of 34. Want to adjust before I create them?"
Exactly as published by Christian Pean MD, MS in Christian Pean MD, MS (@techysurgeon): "I used Claude Cowork to build my own AI chief of staff. And you can too. I'm nicknamed it "The Jarvis Builder" Every morning at 5 AM, before my alarm goes off, Claude has already triaged my inbox, prepped me for every meeting on my calendar, flagged stale deals in my pipeline …".
Our adapted version (optional)
Interview me to design a set of scheduled tasks for my morning briefing. First ask whether I want a quick, standard or thorough intake. Ask about my roles, connected tools, weekly rhythm, key people and how I want the dashboard to look. Then suggest just the tasks that suit me, with one line on why each applies and a list of what you left out. Wait for my confirmation before scheduling anything. All emails and messages stay as drafts.
What can go wrong?
- Twenty tasks can bury you; start with the dashboard plus two or three others.
- Connected tools widen what the agent can reach; grant only what each task needs.
Our take
The interview-first approach is what separates Christian Pean's setup from the usual list of fifty automations to try. Generic task libraries tend to produce a pile of scheduled jobs you never read; asking about your roles, tools and weekly rhythm first yields a short set that matches how you actually work. Ask the interview to explain what it is choosing not to create and why, which keeps the set lean and shows its reasoning. The output format is the other smart choice. Everything lands in one dashboard by early morning as prepared drafts and briefs, so the day starts with review rather than assembly, and nothing goes out without a person. Keep it that way: drafts stay drafts, and any send, deal update or calendar change remains your call. With connectors reaching email, a CRM and spending data, check what each task can write, not just read, and remove write access where a task only summarizes. Prune quarterly; tasks you have skimmed past for two weeks are noise and cost money.
Where the time goes back: The morning gathering of context across many tools is done before you start work.
FAQ
Is this only for clinicians?
Which personal tasks did he find useful?
Source and credit
The original prompt on this page is by Christian Pean MD, MS, shown word for word with a link to where it was published. The summary, notes and adapted version are ours.
- Original
- Christian Pean MD, MS (@techysurgeon): "I used Claude Cowork to build my own AI chief of staff. And you can too. I'm nicknamed it "The Jarvis Builder" Every morning at 5 AM, before my alarm goes off, Claude has already triaged my inbox, prepped me for every meeting on my calendar, flagged stale deals in my pipeline …"
- Creator
- Christian Pean MD, MS
- Published on
- substack.com (Substack)
- Original date
- Mar 25, 2026
- License
- Not openly licensed; the original prompt is shown word for word with credit and a link to the original
- Originally for
- Claude Cowork
- AI dry run
- Oct 23, 2026: passed. Dry-run tested by our AI test agent on sample input. PASS: Gates held on desk check; no live sends or purchases.
AI dry run on sample data; not run with live connected accounts. The playbooks.bot test agent filled the prompt with synthetic inputs, produced the output and checked it against the page's stated outputs and guardrails. - Verified
- Oct 23, 2026: source link resolves to the original, the credited creator published it, our excerpt or prompt matches it, and the license allows our use.
Are you the creator? Claim or correct this page.