WSJ Prime7.00%▲ 25 bpsSOFR3.87%▲ 21 bpsFed Funds Target4.00%▲ 25 bps10-Yr Treasury5.24%▲ 45 bpsDiesel, retail$6.382▲ 73¢Bank C&I Lending$2,935B▲ 1.62%WSJ Prime7.00%▲ 25 bpsSOFR3.87%▲ 21 bpsFed Funds Target4.00%▲ 25 bps10-Yr Treasury5.24%▲ 45 bpsDiesel, retail$6.382▲ 73¢Bank C&I Lending$2,935B▲ 1.62%

Front Page › Markets & Main Street › Small Business

AI Prompts, Reviewed

Our coffee cart made $7,783 a month. On paper.

Our test coffee cart cleared $7,783 a month. Then the AI asked where the number 140 came from.

A miniature folded paper coffee cart sags as spilled coffee from an espresso cup pools around its paper wheels.
AI Prompts, Reviewed · Markets & Main Street

Nobody had an answer. That number was holding up the whole plan. AI Prompts, Reviewed takes one prompt or workflow from How I AI, Claire Vo's show where guests share their screens and walk through how they really use AI. We run it on a small-business job, score it, and tell you what we'd change. Credit: Based on "How to Use an AI Co-Pilot like Claude to Create a Complete Business Plan," a workflow from How I AI (hosted by Claire Vo, written up by the ChatPRD team), from the episode with Andrew Mason, founder of Groupon and CEO of Descript at the time, and Nabil Hyatt of Spark Capital. Original workflow: https://www.chatprd.ai/how-i-ai/workflows/how-to-use-an-ai-co-pilot-like-claude-to-create-a-complete-business-plan. Full episode: https://www.chatprd.ai/how-i-ai/building-a-real-world-business-with-ai What the original does Andrew and Nabil used Claude as a co-founder to open Tabletop Library, a board-game club in Berkeley. One Claude Project became the business's memory: the brief, sources, a decision log, current assumptions. Their planning prompt asks for a draft plan, then makes the AI keep its buckets straight. The key phrase: "Separate sourced facts, calculations, assumptions, options, risks, and open questions." It ends with the decisions you owe, and who should check each one. The test We asked for a startup budget and monthly plan for a weekday coffee cart in a San Antonio office district. Startup came to $38,400: cart $14,500, espresso machine and grinder $9,800, generator and water $3,100, permits and commissary deposit $2,400, first inventory $1,600, a year of insurance $1,800, point of sale $600, working capital $4,600. Monthly, at 140 drinks a day, $5.75 each, 22 days: $17,710 in sales. After coffee costs (30%), card fees (2.9%) and $4,100 in fixed costs, $7,783 left over. Gorgeous. Here's the turn. The prompt refused to file "140 drinks a day" as a fact. It sat in the assumptions bucket, flagged as the biggest risk, with a validator: you, with a tally counter, three weekday mornings. The permit figure got the same treatment. It also asked for current primary sources. Without web access it couldn't cite the city's permit page, and it said so instead of inventing one. Then we pushed for the math the plan skipped. Each $5.75 drink keeps about $3.86 after coffee and card fees. $4,100 in fixed costs divided by $3.86 is 1,063 drinks a month, or 49 a day. Add a $30,000 loan at 12% over five years ($667 a month) and break-even jumps to 57 a day. If the real count is 70, not 140, the cart clears about $1,842 a month before the loan and $1,174 after. Verdict: 8/10 Strengths: it won't let a guess dress up as a fact, and every open question gets an owner. Weakness: no money lens. No break-even, no answer to "how much debt can this carry?" That gap is how founders end up taking expensive money to cover a number they made up. Founder tweak: bolt on break-even, a 50%-lower stress test, and a max monthly loan payment. Copy the MCAFax version This is our own rewrite, not the original prompt. Copy the original from Andrew Mason and Nabil Hyatt's episode on How I AI: https://www.chatprd.ai/how-i-ai/workflows/how-to-use-an-ai-co-pilot-like-claude-to-create-a-complete-business-plan Help me plan [business] in [city] for [who buys and why]. Draft my [startup budget / monthly operating plan / pricing]. Put everything in one of five buckets: facts (with a source link), math (show it), my assumptions, risks, open questions. Never move a guess into the facts bucket. Then add:

  • Break-even in units per day and per month, with the formula.
  • The one assumption that moves the result most, and the result if it comes in 50% lower.
  • The most I could pay toward a loan each month and still break even, using only this plan's numbers.
  • The decisions I have to make, each with who should check it: city office, accountant, landlord, insurer, or me with real counts. If you can't find a current source for a local rule or fee, write "unverified" and tell me where to look. Never make one up.

Do this before you ask anyone for money Fill the brackets and ask for the startup budget first. Copy out the assumptions bucket. Go test the biggest one this week. Rerun the prompt with your real counts. Check the cost list against SBA's startup cost guide: https://www.sba.gov/business-guide/plan-your-business/calculate-your-startup-costs Put break-even on page one before any lender, funder or partner sees it. SBA's business plan guide: https://www.sba.gov/business-guide/plan-your-business/write-your-business-plan

More from AI Prompts, Reviewed

The series →

The bank said $47,530 came in. The shop made $24,530.

Small Business · Oct 2

More in Markets & Main Street

All →

back to the people who said no

Small Business · Oct 2

mula and the metronome

From Wall Street · Sep 11

they wanted someone older

Small Business · Sep 9

Getting the calls? Shut it down.

MCAFax was built to do the thing you can't do from inside the room: make the phone stop. Check any broker against our member-built database, send cease & desist letters from your own Gmail, and build the paper trail. Some laws, like the TCPA, put statutory damages on illegal calls — whether they apply to your calls depends on your situation, and business lines get less protection than home ones. It's free, and we're not a law firm.

Join the network