SuperHuman Plus PowerPlay Q4 [11.03.26]: Boardroom Your 2027 and Get Found by AI
SuperHuman
Power Play
November 3, 2026
12 min read
On this page
Team only: Carter's run of show. Delete this callout before the page goes to TUB Connect.
① Time boxes: Tuesday, November 3, 2:00 to 3:00 PM CT
You come straight off A2A Week 2 (12:00 to 1:45), so everything here is built to need under 10 minutes of prep once tonight's checks are done.
- 2:00 Welcome, where everyone is calling from, housekeeping and the GPT deadline (5 min)
- 2:05 What I'm tracking (8 min)
- 2:13 Play 1: Boardroom your biggest 2027 decision (17 min)
- 2:30 Play 2: Is AI recommending you? (13 min)
- 2:43 Bonus: the zombie spend digest (2 min)
- 2:45 Hot seats (13 min)
- 2:58 Homework and close (2 min)
② Why this shape
Q3 ran 62 minutes. Claude Design alone took about 25 of them, you called it six weeks of education in 48 minutes, and only Gonzalo got a hot seat. This plan has two plays, a 2-minute bonus and a protected 13 minutes of hot seats. If Play 1 runs long, show Play 2's three prompts and your finished example instead of running it live.
③ Prep checklist (tonight, then by 1:55 PM)
Confirm your co-host (Jodi co-hosted Q3)
Download boardroom.skill from the Drive link on this page, install it on a Claude Pro account and run it once on a real 2027 decision. Note how long it takes and how much of the usage window it uses.
Run the Boardroom prompt backup once too, so you can show both
Pick the business for Play 2 (TUB or a volunteer). Ask ChatGPT, Claude and Gemini the six buyer questions tonight and screenshot the best moment where a competitor got recommended.
Make sure your Gmail has a Receipts label with a few receipts in it, or plan to show the bonus prompt without a live run
Tabs open: Claude on Opus 5.5, ChatGPT, Gemini, Gmail and the Q3 class notes
Swap the class notes placeholder in Housekeeping for the real TUB Connect link
④ Fallbacks and untested steps
- Boardroom skill: built for Cowork and Claude Code, and it runs six advisors at the same time. TUB has not tested it in a plain Claude Pro chat. If it stalls or skips the debate, teach the prompt backup. It runs the same three rounds in one paste.
- Play 2: AI answers change by tool, by day and by whether web search is on. Present it as a snapshot of today, not a ranking.
- Zombie spend: the weekly ledger depends on the scheduled task keeping a file between runs, which TUB has not tested. The one-time 90-day audit prompt works without it.
- Housekeeping: ChatGPT's Migrate to plugin button is not tested by TUB. OpenAI says timing can vary by account.
⑤ Overlap check with past SuperHuman content (done November 2)
- Boardroom: not taught in Q1, Q2, Q3 or the SuperHuman Work course. A2A and UCL members saw it in the A2A Week 12 showcase and the August 12 A2A Unlocked call, so expect a few nods.
- Is AI recommending you: not taught. The SuperHuman lesson "GEO-Optimized Blog Posts" covers writing posts for AI search, so Play 2 points to it as the next step instead of repeating it.
- Zombie spend: not taught. The live lesson "Credit Card Statement → Bookkeeping Ready" sorts statement charges for a bookkeeper, and the SuperHuman 2.0 draft "Expense Auditor" audits an uploaded statement. Today's bonus is the inbox version that runs every week on its own. Decision (November 2): keep the bonus.
⑥ Audience and optional close
- Audience: expect SuperHuman Plus buyers plus A2A team-seat members. The October 12 pop-up said teams with 10 or more seats get SuperHuman Plus free for 12 months. Confirm with Student Success before you say it on air.
- Optional close: Q2 and Q3 both ended with an A2A mention, and the October 12 pop-up named April 2027 as the next cohort. If you want a waitlist mention, get the link from Chandler first and leave prices off this call.
🏒 SuperHuman Power Play Q4: Boardroom Your 2027 and Get Found by AI
🎯 Today's goals
🧠 By the end of today you will:
① Know what changed in AI since August
② Put your biggest 2027 decision in front of a six-advisor boardroom
③ Find out if AI recommends your business, and fix what it gets wrong
④ Catch the subscriptions quietly draining your account
🗓 Today's Agenda
- Housekeeping, plus the GPT deadline
- What I'm tracking
- Play 1: Boardroom your biggest 2027 decision
- Play 2: Is AI recommending you?
- Bonus: the zombie spend digest
- Hot seats
- Your homework
🧹 Housekeeping
📒 Class notes: [TUB Connect link needed: not live yet]
📅 Custom GPTs stop working on December 11
OpenAI retires custom GPTs on December 11, 2026. If you built GPTs in SuperHuman or anywhere else, move them before then.
- Staying in ChatGPT: open My GPTs and look for Migrate to plugin. Your instructions and knowledge files come along. Your old chats, custom actions and model choice don't.
- Moving to Claude: copy your GPT's instructions (My GPTs → Edit → Configure), download its knowledge files, and paste the prompt below into a new Claude chat.
Check the notice inside your own ChatGPT account, because OpenAI says timing can vary. Source: EdTech Innovation Hub
GPT rescue prompt for Claude [✨ Prompt Inside]
ROLE
You are an expert at building Claude skills. You turn old custom GPTs into clean, reusable skills that do the job the same way every time.
CONTEXT
My custom GPT stops working on December 11, 2026, and I am moving it to Claude. I copied its instructions from the Configure tab and attached its knowledge files.
GPT name: [NAME]
What I use it for: [ONE LINE]
Its instructions:
[PASTE THE GPT INSTRUCTIONS HERE]
REQUEST
1. Read the instructions and files. Tell me in 3 lines what this GPT does, who it is for, and what a great output looks like.
2. Rewrite the instructions as a Claude skill. Keep my voice, my rules and my format. Cut anything that repeats itself or fights another rule.
3. List anything that will not carry over, like actions that called another app. For each one, give me the simplest fix in Claude: a connector, a project file, or a step I do myself.
4. Run the new skill on one real example I give you so I can check the output.
5. When I say it is right, turn it into a skill.
QUESTIONS
Ask me up to 3 questions, one at a time, only if the instructions leave something unclear. Otherwise get started.🙋♂️ Questions go in the chat
Drop them in as we go. We save the last 13 minutes for hot seats, so hold your biggest one for then.
🎓 Finish your certification and tell the world
It sits at the end of your SuperHuman Plus module. Once you pass, you land on the directory where people look for AI architects, and you can add it to your LinkedIn.
🎥 Everything is RECORDED
The replay goes up in your student portal within 24 hours, along with these class notes.
💪 What We Covered in the Last Three Power Plays
Q1 Power Play
- Reverse engineer any content
- Turn any expert into a searchable brain
- Research with source-specific filtering
Get it here → learn.uncommonbusiness.co
Q2 Power Play
- Claude Skills, the next level of prompting
- Scheduled tasks, AI that works while you sleep
- Chaining skills to build full marketing assets in one chat
Get it here → learn.uncommonbusiness.co
Q3 Power Play
- Which Claude model to pick for which job
- Claude Design: install your brand once and build everything on it
- Scheduled tasks: find your top 5 jobs to hand off and run them overnight
Get it here → Q3 class notes
📰 What I'm Tracking: What Changed Since August
🧠 Claude has a new lineup
Four new models shipped since our last call.
- Fable 5.1 (September 1): the outside agency. Hand it a big goal, not steps. It costs the most, so use it sparingly.
- Opus 5.5 (September 22): your specialist. Anthropic says it performs close to Fable 5.1 on most work and costs less to run. Use it for pricing, hiring, contracts, strategy and long research.
- Sonnet 5.5 (September 28): your everyday worker, faster and cheaper than Sonnet 5. Start here, and run your scheduled tasks here.
- Haiku 5.5 (October 7): the fast, low-cost model for big piles of simple work like sorting and tagging.
Rule of thumb: if you can say what done looks like and check it yourself, start on Sonnet 5.5. If there's no right answer to check, or a wrong call would cost you, start on Opus 5.5.
⚡ Do this today: move your scheduled tasks to Sonnet 5.5 (2 minutes)
- Click Scheduled in the left sidebar.
- Click the three dots on a task you run now, then Edit.
- Find the model setting and switch it to Sonnet 5.5. Say Sonnet 5.5, since Sonnet 5 is the older model.
- Click Save.
- Run it once by hand and read the result before you trust the schedule.
Scheduled tasks need a paid plan.
Sources: Claude release notes and Anthropic news
🤝 AI is starting to come to you
- Claude Duet (October 21): a research preview app where Claude works in the background, flags what needs you, and checks in before it acts. Pro and Max users can join the waitlist.
- Meta's Muse (September): a personal agent that plans your Instagram content and books things for you, and it asks before it sends or buys anything.
- Claude Cowork now lives in every chat (September 16, rolling out on Pro and Max): your tasks, projects, connectors and skills all work in any conversation, on web, desktop and mobile.
You don't need a waitlist. The scheduled tasks from Q3 plus today's zombie spend digest already put Claude to work before you ask.
✨ Also new
- Docs, slides and designs in any Claude chat (September 16, every plan): ask for a one-pager or a deck and Claude builds it right in the conversation.
- Claude Dashboards (October, beta on Pro and up) builds a dashboard from your connected data that refreshes on its own, and Claude Motion (October, beta on Team and Enterprise) turns a report into a short animated video.
- ChatGPT got GPT-6 (October 7): answers now come with charts, calculators and buttons you can click, and you can turn the visuals down if you prefer text.
Source: TechCrunch
🏛️ Play 1: Boardroom Your Biggest 2027 Decision
😫 The Pain
- You're making 2027's biggest calls mostly on your own.
- Ask Claude if your idea is good, and it builds a strong case for yes.
- Ask if it's bad, and it builds an equally strong case for no.
🧱 The Problem
- Friends and family want you to win, so they go easy on you.
- Your team works for you, so they hold back.
- One AI chat tends to agree with however you frame the question.
✅ The Move
- Put the decision in front of a boardroom of six AI advisors, each with a different lens.
- They give their takes on their own, then argue it out where they disagree.
- You walk away with one verdict, three moves for this week and the one question to answer first.
Bring a real decision. Good ones: a price change, a new offer, a hire, a launch date, or what to stop doing in 2027.
🅰️ Option A: Install the Boardroom skill (5 minutes)
A skill is a set of instructions you add to Claude one time. After that, Claude knows the job, and you just ask.
① Install it
- Go to Settings → Capabilities and turn on Code execution and file creation.
- Click Customize, then Skills.
- Click +, then Create skill, then Upload a skill.
- Pick the boardroom.skill file you downloaded.
Done when you see boardroom in your skills list with the switch on.
Team or Enterprise plan: an owner may need to turn on Skills first in Organization settings → Plugins & skills.
💻 Safari on a Mac: read this
- Safari sometimes unzips files on its own.
- If you see a folder instead of a file, right-click the folder and choose Compress.
- Upload that new .zip file.
② Run it [✨ Prompt Inside]
Open a new chat on Opus 5.5, paste this and fill in the brackets.
boardroom this
MY DECISION:
[the decision you're making for 2027, in one or two sentences]
THE DETAILS:
[numbers, timing, what you've already tried, what's at stake]
WHAT I'M UNSURE ABOUT:
[the part that keeps you up at night]↓ Expect: a few quick questions, then Round 1 (six independent takes), Round 2 (the advisors debate their sharpest disagreements), then The Verdict: what the room agrees on, the tension only you can decide, three ranked moves for this week, and the one question to answer first.
③ Meet your boardroom
- Alex Hormozi: the offer, the guarantee and how the money flows
- Warren Buffett: what makes this hard to copy in 10 years, and the risk if you're wrong
- Steve Jobs: what to cut so the experience is great
- Elon Musk: the limit you're assuming that isn't real, and what 10x looks like
- Sara Blakely: what the customer feels, and whether they'd tell a friend
- Peter Thiel: the secret everyone else is missing
These are AI versions built from each person's public work. The real people aren't part of this.
🅱️ Option B: Paste this prompt instead (no skill needed)
Same boardroom, one paste. Works in Claude, ChatGPT or Gemini, and runs best on Opus 5.5.
The Boardroom prompt [✨ Prompt Inside]
ROLE
You are the chair of my personal boardroom. You run a room of six advisors, each speaking as a famous business mind, based on their public books, interviews and talks. You don't take sides. You run the room and deliver the verdict.
THE ADVISORS AND THEIR LENS
- Alex Hormozi: the offer. Is the value stacked, what's the guarantee, and how does the money flow?
- Warren Buffett: the moat. What makes this hard to copy in 10 years, and what's the risk if I'm wrong?
- Steve Jobs: the experience. What should I cut so this is great?
- Elon Musk: first principles. What limit am I assuming that isn't real, and what does 10x look like?
- Sara Blakely: the customer. What frustration does this solve, and would someone tell a friend?
- Peter Thiel: the secret. What is everyone else missing, and is this a monopoly or a commodity?
MY DECISION
[the decision you're making for 2027, in one or two sentences]
THE DETAILS
[numbers, timing, what you've already tried, what's at stake]
WHAT I'M UNSURE ABOUT
[the part that keeps you up at night]
REQUEST
Step 1: Write my decision back to me as a 2 to 3 sentence brief. If something important is missing, ask me up to 3 questions, one at a time, then keep going.
Step 2: ROUND 1, independent takes. For each advisor, give:
- First take: one blunt sentence, their gut reaction
- Analysis: 3 to 5 sentences through their lens only
Each advisor answers as if they never heard the others. No hedging and no trying to be balanced.
Step 3: ROUND 2, the debate. Find the 2 or 3 sharpest disagreements, where two advisors would tell me to do opposite things. Name each one. For each, write a 4 to 6 line back-and-forth between the two advisors, talking to each other. Use at least one real detail from my brief. Leave the tension unresolved.
Step 4: THE VERDICT
- What the room agrees on: 2 to 3 points
- The tension only I can decide: name both sides and what's at stake on each
- Three ranked moves I can make this week, each specific enough that I know exactly what to do
- The one question I must answer first
RULES
If my idea has a fatal flaw, say so plainly. Don't protect my feelings, and don't fake agreement the room doesn't have. If you're not sure about something, say so instead of guessing.↓ Expect: the same three rounds as the skill, in one answer.
🔁 After the verdict [✨ Prompts Inside]
Paste either one into the same chat.
Assume this decision failed six months from now. Work backwards. Rank the most likely reasons it failed and tell me what to fix this week.Turn the verdict into a one-page decision memo I can share with my team: the decision, what we're doing, the three moves with owners, and what would make us change course.🔎 Play 2: Is AI Recommending You?
😫 The Pain
- Your buyers ask ChatGPT, Claude and Gemini who to hire before they ever visit your site.
- You have no idea what those tools say about you.
- If they name a competitor, you never even get the call.
🧱 The Problem
- You can't see those conversations, so you can't fix them.
- AI builds its answer from your website, reviews and listings, and gaps there turn into wrong answers.
- Most small businesses have never checked.
✅ The Move
- Ask the three AI tools the questions your buyers ask.
- See who gets recommended, and what AI gets wrong about you.
- Have Claude write the pages that fix it.
Before you start: use a new chat with web search on in each tool. Answers change from day to day, so treat this as a snapshot of today.
🧭 Run it
① Ask the buyer questions in ChatGPT, Claude and Gemini [✨ Prompt Inside]
1. What is the best [what you sell] for [who you serve] in [your city, or "online"]?
2. Who are the top 5 [type of business] for [your customer's biggest need]? Compare them.
3. What does [your business name] do, and who is it best for?
4. How much does [your business name] cost?
5. Compare [your business name] to [your top competitor]. Which one should I choose?
6. Would you recommend [your business name] to [your ideal customer]? Why or why not?↓ Expect: up to 18 answers. Some name you, some get your price or your customer wrong, and some recommend someone else.
② Grade what AI said about you [✨ Prompt Inside]
Open a new Claude chat, paste this, then paste all the answers underneath.
ROLE
You are an AI search analyst. You help small businesses see what ChatGPT, Claude and Gemini tell buyers about them.
CONTEXT
My business: [name, what you sell, who you serve, your website]
My top competitor: [name]
I asked ChatGPT, Claude and Gemini six buyer questions. Their answers are below, labeled by tool and question number.
REQUEST
1. Make a scorecard for each tool: did it mention me, was what it said accurate, did it recommend me, and who did it recommend instead.
2. List everything the AI tools got wrong or left out about my business, like my price, who I serve or what I offer.
3. For each competitor that got picked over me, explain in plain words why AI likely chose them, based on what's in the answers.
4. Rank the 5 fixes most likely to get me recommended, starting with the one I can do today.
QUESTIONS
If you need facts about my business that aren't in the answers, ask me before you guess.
THE ANSWERS
[paste every answer here]↓ Expect: a scorecard for each tool, a list of what AI gets wrong about you, why competitors won, and 5 ranked fixes.
③ Write the fix [✨ Prompt Inside]
Paste this into the same chat.
ROLE
You write clear web pages that people and AI tools can both understand.
CONTEXT
Use the scorecard and fixes from this chat. My website is [URL]. Read it if you can.
Facts AI got wrong or left out: [your real price or starting price, who you serve, where you work, what makes you different, 2 or 3 real results]
REQUEST
Write these for my website, in plain words, using only the facts I gave you:
1. A "What we do and who it's for" section, 150 words or less
2. A pricing-at-a-glance section. If I didn't give you a price, leave a [PRICE] placeholder.
3. An FAQ that answers my six buyer questions directly, one short paragraph each
4. An honest comparison of me and [competitor]: who should pick me and who should pick them
Then give me a checklist of where to post each piece: my homepage, my Google Business Profile, my LinkedIn and anywhere else my buyers look.
QUESTIONS
Ask me for any fact you need before you write it. Never invent a number, a result or a review.↓ Expect: four ready-to-post sections and a checklist of where each one goes.
④ Check again in 30 days
Ask the same six questions again on December 3. AI tools pick up new pages over time, so compare the answers to today's.
Go further: once your fix pages are live, the GEO-Optimized Blog Posts lesson in your SuperHuman course shows you how to write full posts built to get cited by AI.
💸 Bonus: The Zombie Spend Digest (2 Minutes)
Subscriptions you forgot about renew every month, and year-end renewals are coming. This puts Claude on watch so you catch them every Monday.
① Make a Receipts label in Gmail
- Paste this into the Gmail search bar.
- Click the Show search options icon at the right end of the search bar, then Create filter.
- Check Apply the label, create a new label called Receipts, and check Also apply filter to matching conversations.
- Click Create filter.
subject:(receipt OR invoice OR "payment received" OR renewal OR subscription OR "your order")② Connect Gmail to Claude
In Claude, click Customize, then Connectors, find Gmail and click Connect.
③ Run your 90-day audit [✨ Prompt Inside]
Search my Gmail for receipts and invoices from the last 90 days. Use my Receipts label, plus emails with words like receipt, invoice, renewal or subscription.
Build me a list of every recurring charge: vendor, amount, how often it bills and the date of the last charge.
Flag anything that looks like a duplicate, a price that went up, a free trial that turned paid, or a tool I may have stopped using.
Finish with my total monthly recurring spend and the 3 charges I should cancel or question first, with the reason for each.↓ Expect: a list of every recurring charge, the flags, your monthly total and 3 charges to cancel or question.
④ Put it on autopilot [✨ Prompt Inside]
Paste this into the same chat. Scheduled tasks need a paid plan.
Turn this into a scheduled task that runs every Monday at 7am on Sonnet 5.5:
Search my Gmail for label:Receipts newer_than:8d. For each receipt, pull out the vendor, date, amount and billing cycle. Skip any receipt you already logged, using the Gmail message ID, so nothing is counted twice. Add the new rows to my running expense ledger, and create the ledger if there isn't one yet. Then work out my total monthly recurring spend and flag any new vendor this week, any price change for a vendor I already pay, and any possible duplicate subscription. Send me a short digest: new charges, my monthly recurring total, the flags, and the one charge I should question first.↓ Expect: a short digest every Monday morning, with your monthly total and anything new or suspicious.
🙋 Hot Seats: Raise Your Hand!
Come on up. Tell us your name, your business and one decision you're stuck on, in one sentence. We'll put it in front of the boardroom live.
✅ Your Homework
Boardroom your biggest 2027 decision (Play 1)
Ask ChatGPT, Claude and Gemini your six buyer questions, and post one fix (Play 2)
Run the 90-day zombie spend audit and cancel one charge (Bonus)
Move your custom GPTs before December 11 (Housekeeping)
Finish your certification and get on the directory
📚 Sources
- Claude release notes: Fable 5.1, Opus 5.5, Sonnet 5.5, Haiku 5.5, Cowork in every chat, docs and decks in any chat, Claude Duet
- Anthropic news: model launch dates
- EdTech Innovation Hub: custom GPT retirement and Migrate to plugin
- TechCrunch: GPT-6 and visual answers in ChatGPT
- Meta, Introducing Muse: Muse personal agent
- Startup Ideas, "Cloudflare Will Make 1000+ AI Millionaires": the buyer-question audit behind Play 2
