- Home
- Skills
- Business & Operations
- Medical Business Council
Works with the AI Agents you already use
Medical Business Council
Put a hospital or research-center admin decision before a council of Silicon Valley operators and a compliance Guardian with veto power. Get a scored verdict, the dissent, and a 30-day plan.
$9
Medical Business Council
Example session with this skill installed
Convene the council. Should our medical faculty create a central research support office for grants? Options: A) keep each of our 40 departments handling its own grants, B) a full 8-person office now, C) a 3-person desk piloted in 5 departments. Public university in Egypt, about EGP 25M a year in grants, grant purchases take a median 60 days. Success is doubling grant income in 3 years. I lean toward B.
- Read your context and instructions
- Compiled the medical business council
Verdict: GO on option D (added by the Chair during the debate)
Option scores: A 2.2 · B 3.6 · C 6.8 ·
D 8.4 · Support for the resolution 8.0/10 · Confidence 7/10 · Vote: 7 endorse with amendments, 0 dissent
The call: Don't build the 8-person office yet. Delete the grant-purchase signatures that are internal habit, in all 40 departments, behind a signature map signed by your legal and financial offices. At the same time, pilot a 3-person support desk for 12 months in 5 departments drawn by lot, including 2 that won no grants in the last 3 years.
What carried the room
- Bezos: B is a one-way door. 8 civil-service posts and new signing authority need university council approval and 6 to 12 months of hiring, so it can't deliver a visible change this year.
- Grove: nobody has timed the 60-day purchase chain. Hiring before finding the slowest step just queues faster in front of the same signature.
- Gates and Musk: deleting a habit signature costs nothing per department, so making 35 departments wait a year for a pilot teaches nothing.
Governance Guardian: veto on B until the university council approves the posts and signing changes. No veto on D, as long as only signatures labeled "habit" are deleted and no purchase is split to dodge a threshold.
Next 30 days: time 30 recent purchases, have legal and finance sign the signature map, and open a 30-day objection window for department heads.
Full report: council-sessions/2026-10-31-research-support-office/verdict.md
Connects securely to your tools. The creator never sees your data.
About this skill
Hospitals and research centers lose months to approval chains, slow procurement, and decisions nobody wants to own. This skill puts your hardest non-clinical decision in front of a council of virtual advisors modeled on Jeff Bezos, Andy Grove, Reed Hastings, Satya Nadella, Elon Musk, and Bill Gates, plus a Governance Guardian who knows that "move fast and break things" breaks licenses in healthcare.
The seats argue across three rounds, change their minds only for better arguments, and vote on a written resolution. You get a scored verdict, the strongest dissent, the compliance requirements, and a 30-day plan with owners, kill criteria, and an approval path.
What makes it different
- Advisors grounded in their own words. Each of the six default seats reads a "brain" built from the person's public letters, essays, talks, interviews, and testimony: 65 sources and 1,739 verbatim quotes in total. Every quote is checked word for word against its source before it ships, and every quote a seat uses in your council is checked again.
- A Guardian with a real veto. A governance and compliance seat checks labor law, procurement, audit, data protection, accreditation, and research governance for your country. Its veto cannot be outvoted. It separates what the law requires from what is only internal habit.
- It fights agreement. If every seat agrees in round one, the Chair treats it as a warning and makes the strongest seats argue the alternative nobody raised.
- Honest about limits. Each advisor's brain says where that person's lens misfires in a public hospital, and the seats say so out loud.
What it does
- Full council: five to seven seats plus the Guardian, three rounds, weighted scoring.
- Quick council (
--quick): three seats plus the Guardian, two rounds, for reversible calls. - One-seat consult:
ask grove <question>gets one advisor's answer. - Build an advisor:
build <name>creates a new evidence-grounded seat from public sources. - Board settings: choose who sits on your default board.
How it works
- Brief. Give the decision, two or three options (including "keep the current way"), your numbers and limits, what success looks like, and which way you lean. The council tests your lean instead of agreeing with it.
- Board. The Chair recommends seats for your decision and weights each one by how much the decision lives in its area.
- Round 1. Every seat gives an opening position with scores for each option, ruin risk, and reversibility.
- Round 2. Seats debate the real disagreements and concede what they must.
- Round 3. Seats vote on a draft resolution: endorse, amend, or dissent.
- Verdict. GO, GO WITH CONDITIONS, PILOT FIRST, RESHAPE, WAIT, or DON'T DO IT, with the math shown, saved to
verdict.md.
Use cases
- Cutting a purchase approval chain or a policy manual that has grown out of control
- Outsourcing billing, maintenance, or another support function
- Setting up a grants office or a research funding scheme
- Rolling out an electronic records system or AI documentation tools
- Staffing models, incentives, reorganization, and leadership changes
- New revenue lines, partnerships, and medical tourism
What's inside
SKILL.md (the council protocol), roster.json (12 seats), six advisor brains, reference files for scoring, round prompts, and the report template, a quote-checking script, the advisor-building pipeline, and a worked example.
Known limitations
- Not for clinical decisions: no diagnosis, treatment, specific patients, clinical protocols, or study design.
- Decision support only, not legal, financial, or regulatory advice. The Guardian names rules to check; confirm them with your legal and compliance offices.
- The advisors are simulations built from public words. They are not those people and are not endorsed by them.
- Works best in agents that can run sub-agents in parallel (such as Claude Code). Other agents run the seats one after another, which is slower.
- A full council uses about 21 sub-agent runs; use
--quick(about 8) for smaller decisions. - The five on-call seats (Jobs, Collison, Doerr, Horowitz, Huang) reason from summary cards until you build their brains.
- The quote check needs Python 3 (standard library only).
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Business & Operations.
No clinical or patient-level advice. Requires Python for quote verification. Admin decisions only.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 6 days ago
- Software
Python 3.10+
- Optional
- Internet accessMarkdown viewerPDF export tool