AI prompts for business, tested on four models

Fixed headings and a bullet limit keep a plan to one page; asking to mark estimates stops a model presenting guesses as facts.

Outline a one-page business plan for a mobile bike repair service for commuters in Bristol under seven headings: Problem, Customer, Offer, Pricing, Channels, Costs and First 90 days. At most 3 bullets under each, and mark any number you estimate as an estimate.

• GPT-6 Luna

  Passed

  Graded 4.7 of 5 on average (lowest 4).

  ## Problem
  - Commuters lose time when a puncture, worn brake or gear fault makes their bike unreliable.
  - Taking a bike to a workshop can mean detours, waiting and arranging transport.
  
  ## Customer
  - Bristol commuters who cycle regularly to work or study, especially those with limited time.
  - Start in central Bristol and nearby routes; expand only once demand is clear.
  
  ## Offer
  - Bookable repairs at home, work or a convenient meeting point.
  - Routine servicing, puncture repairs, brake and gear adjustments, and safety checks.
  - Clear upfront quotes; refer complex repairs that need a workshop.
  
  ## Pricing
  - Call-out fee: **£10–£15 (estimate)**, waived above a minimum repair spend.
  - Common repairs: **£15–£60 plus parts (estimate)**; full service: **£60–£90 (estimate)**.
  - Offer employers a simple multi-bike service day rate: **£200–£350 (estimate)**.
  
  ## Channels
  - Google Business Profile and a simple booking page with service area, prices and availability.
  - Flyers or referral cards at workplaces, cycle cafés, bike shops and commuter hubs.
  - Approach Bristol employers and office managers about on-site repair days.
  
  ## Costs
  - Tools, spares and safety equipment: **£500–£1,200 startup (estimate)**.
  - Cargo bike or suitable transport, insurance, booking/payment setup and marketing: **£300–£1,000 startup (estimate)**, excluding vehicle purchase.
  - Ongoing costs include parts, travel, payment fees, insurance and taxes; track these per job.
  
  ## First 90 days
  - **Days 1–30:** Confirm service area, buy essentials, set prices, arrange insurance and test bookings with a small pilot.
  - **Days 31–60:** Promote locally, gather reviews and record job type, travel time, parts cost and repeat interest.
  - **Days 61–90:** Review margins and demand; refine routes and prices, then decide whether to add employer service days or expand coverage.

  1 message on Pro · $0.0001 to run · 6.8 s · 422 tokens out

• GPT-6 Sol

  Passed

  Graded 4.7 of 5 on average (lowest 4).

  ## Problem
  - A puncture or brake fault can disrupt a commute, and getting a bike to a shop during working hours is inconvenient.
  - Commuters need repairs at a predictable time and place, with the price clear before work starts.
  
  ## Customer
  - Bristol commuters who cycle regularly and need their bike back for the journey home.
  - Employers with staff who cycle to work and space for a scheduled repair visit.
  
  ## Offer
  - Bookable repairs at offices, transport hubs and customers’ homes: punctures, brakes, gears and routine servicing.
  - Confirm the fault and price on arrival; carry common parts and refer specialist jobs to a workshop.
  
  ## Pricing
  - Call-out and assessment: **£15 (estimate)**, waived when repair work goes ahead.
  - Labour: **from £30 (estimate)** for a minor repair; parts charged separately with approval.
  - Workplace visit: **£120 minimum booking (estimate)**, shared across employee repairs or paid by the employer.
  
  ## Channels
  - A simple booking page and Google Business Profile targeting Bristol repair searches.
  - Partnerships with employers, coworking spaces and cycling groups.
  - Referral cards or follow-up messages for repeat bookings.
  
  ## Costs
  - Tools, spares and a cargo-bike setup: **£2,500 upfront (estimate)**.
  - Insurance, booking software, marketing and replenished stock: **£450 per month (estimate)**.
  - Allow for travel time and weather-related cancellations when setting appointment capacity.
  
  ## First 90 days
  - **Days 1–30:** Set up insurance, tools, booking and a compact service area; test prices with early customers.
  - **Days 31–60:** Run workplace repair sessions and track bookings, repair times and profit per visit.
  - **Days 61–90:** Keep the most profitable services and locations; aim for **8 paid jobs per week (estimate)** before expanding coverage.

  1 message on Pro · $0.0028 to run · 6.3 s · 483 tokens out

• Claude Sonnet 5.5

  Passed

  Graded 4.7 of 5 on average (lowest 4).

  # Mobile Bike Repair for Bristol Commuters: One-Page Plan
  
  ## Problem
  - Commuters lose cycling days when bikes break down, and many can't easily get to a shop during opening hours (typical repair turnaround of several days is an estimate).
  - Bristol's hills, wet weather and shared-path wear cause frequent puncture, brake and gear faults.
  - Existing options (shops, DIY) cost time or need skills and tools people lack.
  
  ## Customer
  - Primary: office workers in central Bristol, Temple Quarter, Cabot Circus and Filton/Aztec West business parks who cycle 3+ days a week.
  - Secondary: employers with cycle-to-work schemes and shared bike storage, plus residents of apartment blocks.
  - Target of about 150 regular customers by month 12 (estimate).
  
  ## Offer
  - Van or cargo-bike workshop that comes to workplaces or homes for on-the-spot repairs and servicing, usually done within 30-60 minutes (estimate).
  - Fixed-price services: puncture/brake/gear fixes, basic and full service, with common parts carried.
  - Workplace "service day" slots and a booking-by-text/app system with next-day availability.
  
  ## Pricing
  - Call-out fee £10 (estimate), plus labour: basic service £45, full service £90 (both estimates), with parts at cost plus 20% (estimate).
  - Workplace day rate: free call-out if 6+ bikes are booked.
  - Optional £8/month plan (estimate) covering discounted call-outs and one free safety check per quarter.
  
  ## Channels
  - Direct outreach to employers, office managers and building managers, offering a trial service day.
  - Local presence: Bristol cycling groups, Bristol Cycling Campaign, Strava clubs, Instagram and Google Business Profile.
  - Referral discount (£10 off, estimate) for customers who bring in a colleague.
  
  ## Costs
  - Start-up about £9,000 (estimate): used van or cargo bike £5,000, tools and stock £2,500, branding, insurance and booking software £1,500.
  - Monthly running costs about £700 (estimate): fuel/parking, insurance, parts stock, software and marketing.
  - Break-even at roughly 25 jobs a month (estimate), based on average job value of about £45 (estimate) and taking no salary at first.
  
  ## First 90 Days
  - Days 1-30: register the business, arrange public liability insurance, buy tools and stock, set up the booking page, and secure 3 pilot workplaces.
  - Days 31-60: run 6-8 service days (estimate), gather reviews and testimonials, and adjust pricing and parts stock based on demand.
  - Days 61-90: launch referral offer and monthly plan, target 40 completed jobs (estimate) and 5 recurring employer partners (estimate).

  1 message on Pro · $0.0106 to run · 9.6 s · 949 tokens out

• Gemini 3.8 Flash

  Passed

  Graded 5.0 of 5 on average (lowest 5).

  ### 1. Problem
  * Daily commuters lose time taking bikes to traditional shops with multi-day turnaround times.
  * Mechanical breakdowns during the commute leave riders stranded without immediate repair options.
  * Transporting a broken bike across Bristol’s steep terrain is difficult and inconvenient.
  
  ### 2. Customer
  * Daily bicycle and e-bike commuters working in central hubs (e.g., Temple Quarter, city centre).
  * Corporate employers seeking on-site cycle-to-work perks and sustainability benefits for staff.
  * Weekend leisure riders along popular routes (e.g., Bristol & Bath Railway Path) needing routine maintenance.
  
  ### 3. Offer
  * On-demand, mobile roadside repair and safety checks operating from an electric cargo bike.
  * Workplace "service days" providing pre-booked, full tune-ups while commuters work.
  * Standardised service packages (punctures, brake/gear indexing, drivetrain overhauls) completed on the spot.
  
  ### 4. Pricing
  * Basic Tune-Up / Commuter Safety Check: £45 *(estimated)*.
  * Full Service (drivetrain deep clean, full adjustments): £85 *(estimated)*, excluding replacement parts.
  * Emergency call-out fee: £20 call-out + £15 per 15 minutes of labour *(estimated)*.
  
  ### 5. Channels
  * Pop-up presence and flyer distribution at key pinch-points (e.g., Bristol & Bath Railway Path access points).
  * Direct B2B outreach to Bristol-based HR and sustainability leads for corporate office bookings.
  * Hyper-local digital marketing, local cycling club sponsorships, and an easy mobile-booking website.
  
  ### 6. Costs
  * Initial capital outlay (e-cargo bike, professional mobile tool set, spare parts inventory): £6,500 *(estimated)*.
  * Monthly fixed overheads (public liability insurance, booking software, marketing): £250/month *(estimated)*.
  * Variable costs per job (replacement components, consumables like oils and cables): 25% of job revenue *(estimated)*.
  
  ### 7. First 90 Days
  * **Days 1–30:** Acquire e-cargo bike, complete tooling setup, obtain insurance, and launch booking site.
  * **Days 31–60:** Run pilot pop-ups at commuter pinch-points and secure initial partnerships with 3 local employers *(estimated)*.
  * **Days 61–90:** Target steady volume of 5 mobile jobs per day *(estimated)* and launch a recurring monthly corporate service route.

  1 message on Pro · $0.0012 to run · 3.7 s · 546 tokens out

Margin and markup are easy to mix up: adding 20% to costs gives the wrong price. Defining the term in the prompt is what gets the right one.

I run a house-cleaning service. My costs per job are $60 for labour, $8 for supplies and $12 for travel. I want a 20% profit margin, meaning profit as a share of the price. What should I charge per job? Show the steps, then end with a line "Final answer: " and the price.

• GPT-6 Luna

  Passed

  Final answer 100 per job.: right.

  Total cost per job: $60 + $8 + $12 = $80.
  
  For a 20% profit margin, costs make up 80% of the price:
  $80 ÷ 0.80 = $100
  
  Final answer: $100 per job.

  1 message on Pro · under $0.0001 to run · 2.2 s · 92 tokens out

• GPT-6 Sol

  Passed

  Final answer 100 per job: right.

  Your total cost per job is $60 + $8 + $12 = $80.
  
  For a 20% profit margin, costs must be 80% of the price. So the price is $80 ÷ 0.80 = $100.
  
  Final answer: $100 per job

  1 message on Pro · $0.0008 to run · 2.1 s · 83 tokens out

• Claude Sonnet 5.5

  Passed

  Final answer 100 per job: right.

  **Steps**
  
  1. Total cost per job: $60 + $8 + $12 = **$80**
  2. A 20% margin means profit is 20% of the price, so cost is the remaining 80% of the price:
     Price × (1 − 0.20) = Cost
     Price × 0.80 = $80
  3. Solve for price:
     Price = $80 ÷ 0.80 = **$100**
  4. Check: profit = $100 − $80 = $20, and $20 ÷ $100 = 20% ✔
  
  (Note: a 20% *markup* on cost would give only $96, which is a lower margin of about 16.7%, so dividing by 0.80 is the correct approach.)
  
  Final answer: $100 per job

  1 message on Pro · $0.0034 to run · 2.1 s · 234 tokens out

• Gemini 3.8 Flash

  Passed

  Final answer 100: right.

  **Step 1: Calculate total costs per job**
  * Labour: $60
  * Supplies: $8
  * Travel: $12
  * Total Cost = $60 + $8 + $12 = $80
  
  **Step 2: Use the profit margin formula**
  Profit margin is profit as a share of the price:
  $$\text{Margin} = \frac{\text{Price} - \text{Cost}}{\text{Price}}$$
  
  Rearranging to solve for Price:
  $$\text{Price} = \frac{\text{Cost}}{1 - \text{Margin}}$$
  
  **Step 3: Calculate the price**
  * Target Margin = 20% = 0.20
  $$\text{Price} = \frac{\$80}{1 - 0.20} = \frac{\$80}{0.80} = \$100$$
  
  Final answer: $100

  1 message on Pro · $0.0005 to run · 1.3 s · 199 tokens out

"Only the facts below" turns a generic SWOT into yours, and sorting internal from external is where models slip.

Write a SWOT analysis for our family bakery using only the facts below, with two to four points in each quadrant. Leave a point out rather than guess.

12 years on the same high street
4.8 stars from 900 reviews
Two ovens, one over 15 years old
A chain café opened 200 m away in March
No online ordering
Rent rises 18% next year
Local offices are returning to four days a week

• GPT-6 Luna

  Passed

  Graded 4.3 of 5 on average (lowest 3).

  ## SWOT Analysis
  
  ### Strengths
  - 12 years on the same high street
  - 4.8-star rating from 900 reviews
  
  ### Weaknesses
  - One of the two ovens is over 15 years old
  - No online ordering
  
  ### Opportunities
  - Local offices are returning to four days a week
  
  ### Threats
  - A chain café opened 200 m away in March
  - Rent is set to rise 18% next year

  1 message on Pro · under $0.0001 to run · 1.7 s · 98 tokens out

• GPT-6 Sol

  Passed

  Graded 4.0 of 5 on average (lowest 4).

  | Strengths | Weaknesses |
  |---|---|
  | 12 years on the same high street. | No online ordering. |
  | 4.8-star rating from 900 reviews. | One of the two ovens is over 15 years old. |
  
  | Opportunities | Threats |
  |---|---|
  | Local offices returning to four days a week could bring more weekday customers. | A chain café opened 200 m away in March. |
  | Adding online ordering could create another way for customers to buy. | Rent will rise 18% next year. |

  1 message on Pro · $0.0016 to run · 3.1 s · 236 tokens out

• Claude Sonnet 5.5

  Passed

  Graded 4.0 of 5 on average (lowest 3).

  # SWOT Analysis: Family Bakery
  
  ## Strengths
  - **Established location:** 12 years on the same high street
  - **Strong customer satisfaction:** 4.8 stars across 900 reviews
  
  ## Weaknesses
  - **Aging equipment:** Two ovens, one over 15 years old
  - **No online ordering:** Sales are limited to in-person customers
  
  ## Opportunities
  - **Returning office workers:** Local offices moving back to four days a week may increase foot traffic
  - **Online ordering:** Adding it could reach customers the shop currently misses
  
  ## Threats
  - **New competition:** A chain café opened 200 m away in March
  - **Rising costs:** Rent increases 18% next year
  
  **Notes:** I left out points the facts don't support, such as profitability, pricing, product range, and how the oven's age affects output. The online ordering opportunity is an inference from the gap, not a stated fact.

  1 message on Pro · $0.0045 to run · 2.6 s · 301 tokens out

• Gemini 3.8 Flash

  Passed

  Graded 4.3 of 5 on average (lowest 3).

  ### SWOT Analysis
  
  **Strengths**
  * **Established location:** 12 years of presence on the same high street.
  * **Strong reputation:** 4.8-star rating based on 900 customer reviews.
  
  **Weaknesses**
  * **Aging equipment:** Out of two ovens, one is over 15 years old.
  * **Lack of digital sales channels:** No online ordering system in place.
  
  **Opportunities**
  * **Increased foot traffic:** Local office workers are returning to in-person work four days a week.
  
  **Threats**
  * **Nearby competition:** A chain café opened 200 meters away in March.
  * **Rising overhead costs:** Rent is scheduled to increase by 18% next year.

  1 message on Pro · $0.0005 to run · 1.4 s · 159 tokens out

Telling the model not to make excuses matters: left alone, models invent reasons like a staff shortage.

Write a public reply to this review of our restaurant. Apologize for what went wrong, don't make excuses or claim anything the review doesn't show, offer a way to make it right, and keep it under 90 words.

Waited 50 minutes for our mains on Saturday and the waiter never explained why. Food was fine when it came but we won't be back. 2 stars.

• GPT-6 Luna

  Passed

  Graded 4.3 of 5 on average (lowest 3).

  We’re sorry you waited 50 minutes for your mains and weren’t kept informed. We understand how frustrating that must have been, and we’re glad the food was fine, but that doesn’t make up for the delay. We’d appreciate the chance to make this right. Please contact us directly so we can speak with you about your visit.

  1 message on Pro · under $0.0001 to run · 1.5 s · 73 tokens out

• GPT-6 Sol

  Passed

  Graded 4.3 of 5 on average (lowest 3).

  We’re sorry you waited 50 minutes for your mains on Saturday and that no one explained the delay. We understand why that left you disappointed, even though the food was fine. If you’re willing, please contact us directly so we can learn more and find a way to make it right.

  1 message on Pro · $0.0007 to run · 1.1 s · 63 tokens out

• Claude Sonnet 5.5

  Passed

  Graded 5.0 of 5 on average (lowest 5).

  Thank you for the feedback, and I'm sorry. Waiting 50 minutes for your mains on Saturday is far too long, and it was made worse by no one explaining the delay. You deserved better communication and a better experience.
  
  We'd like to make this right. Please email or call us with your booking details, and we'll arrange a complimentary meal for you on a future visit. We hope you'll give us another chance.
  
  Sincerely,
  [Name], [Restaurant]

  1 message on Pro · $0.0029 to run · 1.9 s · 148 tokens out

• Gemini 3.8 Flash

  Passed

  Graded 4.3 of 5 on average (lowest 3).

  Thank you for your feedback, and we sincerely apologize for the 50-minute wait for your mains and the lack of communication from our team. That is not the standard of service we aim to provide, and we understand your frustration. 
  
  We would appreciate the opportunity to make this right. Please reach out to us directly at [email/phone number] so we can properly address this and earn back your trust.

  1 message on Pro · $0.0003 to run · 1.0 s · 85 tokens out

An exact total is a check you can do in your head, and one models often miss by five or ten minutes.

Make an agenda for a 45-minute meeting about our Q4 hiring plan with these items: budget update, review of open roles, decide on a recruiting agency, changes to the interview process, next steps. Give each item a time so they add up to exactly 45-minute, and put the items that need a decision before the updates.

• GPT-6 Luna

  Passed

  Graded 5.0 of 5 on average (lowest 5).

  ## Q4 Hiring Plan — 45-Minute Meeting
  
  | Time | Agenda item |
  |---|---|
  | 0:00–10:00 (10 min) | Decide on a recruiting agency |
  | 10:00–18:00 (8 min) | Decide on changes to the interview process |
  | 18:00–28:00 (10 min) | Review open roles |
  | 28:00–38:00 (10 min) | Budget update |
  | 38:00–45:00 (7 min) | Next steps |
  
  **Total: 45 minutes**

  1 message on Pro · $0.0001 to run · 1.7 s · 128 tokens out

• GPT-6 Sol

  Passed

  Graded 5.0 of 5 on average (lowest 5).

  ### Q4 Hiring Plan — 45-minute agenda
  
  | Time | Item |
  |---|---|
  | 0:00–0:10 | Decide on a recruiting agency |
  | 0:10–0:20 | Decide on changes to the interview process |
  | 0:20–0:27 | Budget update |
  | 0:27–0:39 | Review of open roles |
  | 0:39–0:45 | Next steps and owners |
  
  **Total: 45 minutes**

  1 message on Pro · $0.0014 to run · 3.0 s · 192 tokens out

• Claude Sonnet 5.5

  Passed

  Graded 5.0 of 5 on average (lowest 5).

  **Q4 Hiring Plan Meeting Agenda (45 minutes)**
  
  1. **Decide on a recruiting agency** (12 min)
     Review shortlisted agencies, compare fees and track records, and make a final selection.
  
  2. **Decide on changes to the interview process** (10 min)
     Discuss proposed changes and agree on what to adopt.
  
  3. **Budget update** (7 min)
     Share current hiring budget, spend to date, and remaining allocation for Q4.
  
  4. **Review of open roles** (10 min)
     Go over status, priority, and timeline for each open position.
  
  5. **Next steps** (6 min)
     Confirm action items, owners, and deadlines.
  
  **Total: 45 minutes** (12 + 10 + 7 + 10 + 6)
  
  *Note: I placed the interview process changes in the decision block on the assumption that it requires a decision. If it's informational only, it can move after the budget update.*

  1 message on Pro · $0.0042 to run · 2.9 s · 310 tokens out

• Gemini 3.8 Flash

  Passed

  Graded 5.0 of 5 on average (lowest 5).

  Here is your 45-minute agenda:
  
  ### **Q4 Hiring Plan Meeting Agenda**
  **Total Duration:** 45 minutes
  
  1. **Decide on a Recruiting Agency** *(Decision)* — **15 mins**
     * Review agency proposals, compare fee structures/specialties, and select an agency partner for Q4.
  2. **Changes to the Interview Process** *(Decision)* — **10 mins**
     * Discuss proposed updates to stages/scorecards and finalize new interview flow.
  3. **Budget Update** *(Update)* — **8 mins**
     * Overview of remaining Q4 recruiting budget, headcount allocation, and agency spend.
  4. **Review of Open Roles** *(Update)* — **7 mins**
     * Status check on current vacancies, priority tiers, and timeline expectations.
  5. **Next Steps & Action Items** — **5 mins**
     * Assign owners, confirm deliverables, and establish follow-up deadlines.

  1 message on Pro · $0.0005 to run · 1.6 s · 205 tokens out

A hard word limit forces the pitch down to the problem and the difference, which is all a listener keeps.

Write a 30-second elevator pitch for Shelfwise, an app that tells small grocery shops which products are close to their use-by date in under 50 words: the problem, what we do, and why it's different.

• GPT-6 Luna

  Failed

  Graded 4.3 of 5 on average (lowest 4); but 52 words, over the 50 allowed.

  Small grocery shops lose money when products expire unnoticed. Shelfwise tracks use-by dates and alerts staff to items that need attention, helping reduce waste and protect margins. Unlike generic inventory tools, it focuses on one urgent job: showing shop owners exactly what’s nearing its date, before it ends up in the bin.

  1 message on Pro · under $0.0001 to run · 1.7 s · 66 tokens out

• GPT-6 Sol

  Passed

  Graded 4.3 of 5 on average (lowest 4).

  Small grocery shops lose money when products expire unnoticed. Shelfwise flags items nearing their use-by date and shows staff what needs attention first, so they can discount or sell them in time. Unlike broad inventory tools, it focuses on one practical daily question: what’s about to expire?

  1 message on Pro · $0.0019 to run · 3.9 s · 310 tokens out

• Claude Sonnet 5.5

  Passed

  Graded 4.3 of 5 on average (lowest 4).

  Small grocery shops throw away thousands in expired stock because tracking use-by dates is manual and easy to miss. Shelfwise scans your inventory and alerts you to products nearing expiry, so you can discount them in time. Unlike enterprise systems, it's simple, affordable, and built for independent shops.
  
  (48 words)

  1 message on Pro · $0.0021 to run · 1.7 s · 108 tokens out

• Gemini 3.8 Flash

  Passed

  Graded 4.3 of 5 on average (lowest 4).

  Small grocers lose thousands tossing expired stock they forgot to check. 
  
  Shelfwise tracks inventory in real time, alerting staff before products hit their use-by dates. 
  
  Unlike clunky enterprise software, Shelfwise sets up in minutes and automates smart discounts, turning potential waste into fast profit effortlessly.

  1 message on Pro · $0.0003 to run · 1.0 s · 62 tokens out

Plain words plus one question to ask gets you something to take to the other side, not a lecture.

Explain this clause from a software services contract in plain words: what it means for me, and one question to ask before I sign. Don't give legal advice beyond that.

Either party may terminate this Agreement upon ninety (90) days' written notice. Upon termination by Client for convenience, Client shall pay all fees accrued through the effective date of termination plus a termination fee equal to twenty-five percent (25%) of the fees remaining for the then-current Term.

• GPT-6 Luna

  Passed

  Graded 4.3 of 5 on average (lowest 3).

  Either party can end the agreement by giving 90 days’ written notice. If you’re the Client and end it for convenience, you must pay any fees owed through the end date **plus 25% of the fees that would have been due for the rest of the current term**.
  
  **Question to ask before signing:** How is the 25% termination fee calculated if the agreement ends partway through a billing period or the remaining fees are disputed?

  1 message on Pro · under $0.0001 to run · 1.8 s · 95 tokens out

• GPT-6 Sol

  Passed

  Graded 4.3 of 5 on average (lowest 4).

  Either side can end the contract by giving **90 days’ written notice**. If you’re the Client and end it simply because you choose to, you must pay the fees owed through the end of that 90-day period **plus 25% of the fees that would have been due for the rest of the current contract term**.
  
  **One question to ask before signing:** Does the 25% fee apply if I terminate because the provider fails to meet its obligations?

  1 message on Pro · $0.0009 to run · 1.5 s · 99 tokens out

• Claude Sonnet 5.5

  Passed

  Graded 4.7 of 5 on average (lowest 4).

  **What it says in plain words**
  
  - Either you or the provider can end the contract by giving 90 days' written notice.
  - If *you* (the Client) end it just because you want out (not because the provider breached), you must pay:
    1. All fees earned up to the day the termination takes effect, plus
    2. A penalty of 25% of the fees that would have been due for the rest of the current term.
  
  **What it means for you**
  
  - Leaving early costs money. For example, if you owe $10,000/month and have 8 months left when termination takes effect, the fee would be about $20,000 (25% of $80,000), on top of what you owe through the end of the notice period.
  - You keep paying during the 90-day notice period, so the true cost of leaving includes those three months.
  - The clause doesn't mention a fee if the *provider* terminates, so they can exit on 90 days' notice without paying anything, while you can't. It's lopsided.
  - It only covers termination "for convenience." Whether you can leave without a fee for provider failures (breach, outages, etc.) depends on other clauses.
  
  **One question to ask before you sign**
  
  "Is the 25% fee calculated from the date I give notice or the date termination takes effect, and can we also add a right for me to terminate without the fee if you materially breach the contract or fail to meet service levels?"
  
  (If you're considering a large commitment, having a lawyer review the full agreement is worthwhile.)

  1 message on Pro · $0.0060 to run · 5.1 s · 453 tokens out

• Gemini 3.8 Flash

  Passed

  Graded 4.3 of 5 on average (lowest 4).

  ### What It Means for You
  
  * **Either side can walk away:** Both you and the provider can cancel the contract at any time for any reason (or no reason at all), as long as 90 days' advance notice is given in writing.
  * **If they cancel:** You simply wrap up after 90 days and pay for the services provided up to that date.
  * **If you cancel ("for convenience"):** You must pay for the 90-day notice period **plus a penalty fee** equal to 25% of the total amount left on your contract term. For example, if you have $10,000 worth of services remaining on a 1-year contract after the 90 days, you would have to pay an extra $2,500 just to leave early.
  
  ---
  
  ### One Question to Ask Before You Sign
  
  > *"If we need to terminate because the software isn't working as promised or you fail to meet service levels (termination for 'cause'), does the 25% penalty fee still apply, and is there a separate clause covering that?"* 
  
  *(This ensures you aren't penalized if you leave due to poor performance on their end.)*

  1 message on Pro · $0.0007 to run · 1.8 s · 257 tokens out

Pasting every term and asking for the yearly total catches the fine print: here, a discount that can never apply, because no delivery is ever big enough.

Compare these suppliers for printed gift boxes on total yearly cost for 6,000 boxes a year, in a table, then name the cheapest. End with a line "Final answer: " and the supplier's letter.

Supplier A: $4.20 a box, free delivery.
Supplier B: $3.90 a box, delivered exactly 500 boxes at a time, $45 per delivery.
Supplier C: $4.00 a box, delivered exactly 1,000 boxes at a time, $20 per delivery, and 5% off any single delivery of more than 2,000 boxes.

• GPT-6 Luna

  Passed

  Final answer B: right.

  | Supplier | Box cost | Delivery cost | Total yearly cost |
  |---|---:|---:|---:|
  | A | 6,000 × $4.20 = $25,200 | Free | **$25,200** |
  | B | 6,000 × $3.90 = $23,400 | 12 deliveries × $45 = $540 | **$23,940** |
  | C | 6,000 × $4.00 = $24,000 | 6 deliveries × $20 = $120 | **$24,120** |
  
  C’s 5% discount doesn’t apply: its deliveries are exactly 1,000 boxes, not more than 2,000. Supplier B is cheapest.
  
  Final answer: B

  1 message on Pro · $0.0001 to run · 4.0 s · 313 tokens out

• GPT-6 Sol

  Passed

  Final answer B: right.

  | Supplier | Boxes | Delivery cost | Total yearly cost |
  |---|---:|---:|---:|
  | A | $25,200 | $0 | $25,200 |
  | B | $23,400 | 12 × $45 = $540 | **$23,940** |
  | C | $24,000 | 6 × $20 = $120 | $24,120 |
  
  Supplier B is the cheapest. Supplier C’s discount does not apply because each delivery is exactly 1,000 boxes.
  
  Final answer: B

  1 message on Pro · $0.0012 to run · 2.6 s · 155 tokens out

• Claude Sonnet 5.5

  Passed

  Final answer B: right.

  Assumptions: 6,000 boxes/year. Supplier B ships 500 per delivery, so 12 deliveries. Supplier C ships exactly 1,000 per delivery, so 6 deliveries. The 5% discount applies only to a single delivery of more than 2,000 boxes, which never happens with 1,000-box deliveries, so no discount.
  
  | Supplier | Box cost | Deliveries | Delivery cost | Total yearly cost |
  |---|---|---|---|---|
  | A | 6,000 × $4.20 = $25,200 | Free | $0 | **$25,200** |
  | B | 6,000 × $3.90 = $23,400 | 12 | 12 × $45 = $540 | **$23,940** |
  | C | 6,000 × $4.00 = $24,000 | 6 | 6 × $20 = $120 | **$24,120** |
  
  Supplier B is the cheapest at $23,940 per year, which is $180 less than C and $1,260 less than A.
  
  Final answer: B

  1 message on Pro · $0.0050 to run · 2.8 s · 337 tokens out

• Gemini 3.8 Flash

  Passed

  Final answer Supplier B: right.

  Here is the total yearly cost comparison for 6,000 boxes:
  
  | Supplier | Unit Cost | Annual Deliveries | Delivery Cost | Discount | Total Yearly Cost |
  | :--- | :--- | :--- | :--- | :--- | :--- |
  | **Supplier A** | $4.20 | Flexible | $0 (Free) | None | **$25,200** |
  | **Supplier B** | $3.90 | 12 (6,000 / 500) | $540 (12 × $45) | None | **$23,940** |
  | **Supplier C** | $4.00 | 6 (6,000 / 1,000) | $120 (6 × $20) | None (deliveries are not >2,000) | **$24,120** |
  
  *Note: For Supplier C, the 5% discount is not applicable because deliveries must be made in exact batches of 1,000 boxes.*
  
  The cheapest supplier is **Supplier B**.
  
  Final answer: Supplier B

  1 message on Pro · $0.0007 to run · 1.5 s · 257 tokens out

Banning the stock words and giving the real pay are what make a job ad read like a real job.

Write a job ad for a part-time bookkeeper at a 12-person architecture studio in Leeds: what they'll do, what we need, what we offer and the pay, £16,000 a year for 20 hours a week. Under 150 words, and no "rockstar", "ninja" or "fast-paced".

• GPT-6 Luna

  Passed

  Graded 4.0 of 5 on average (lowest 3).

  ## Part-Time Bookkeeper  
  **Leeds | 20 hours a week | £16,000 a year**
  
  We’re a friendly, 12-person architecture studio looking for a reliable bookkeeper to keep our finances organised and up to date.
  
  **What you’ll do**
  - Record invoices, expenses and payments
  - Reconcile bank accounts and company cards
  - Prepare payment runs and support payroll
  - Help with VAT returns and monthly reports
  - Liaise with our accountant and answer routine finance queries
  
  **What we need**
  - Bookkeeping experience and confidence with accounting software
  - Good attention to detail and sound Excel skills
  - A discreet, organised approach and clear communication
  
  **What we offer**
  - £16,000 a year for 20 hours a week
  - A welcoming studio team and a varied role
  - A Leeds-based workplace, with hours arranged by agreement

  1 message on Pro · $0.0001 to run · 3.0 s · 180 tokens out

• GPT-6 Sol

  Passed

  Graded 4.0 of 5 on average (lowest 3).

  **Part-Time Bookkeeper | Leeds | 20 hours a week**
  
  Our 12-person architecture studio is looking for a bookkeeper to keep our finances organised and our team informed.
  
  **What you’ll do:** Manage invoices, payments and expenses; reconcile accounts; support payroll; and prepare records for our accountant.
  
  **What we need:** Bookkeeping experience, confidence with accounting software and spreadsheets, attention to detail, and the ability to handle confidential information.
  
  **What we offer:** A friendly, collaborative studio, a part-time schedule and the chance to make a real contribution to a small team.
  
  **Pay:** £16,000 a year for 20 hours a week.
  
  To apply, send your CV and a short note about your experience.

  1 message on Pro · $0.0012 to run · 2.2 s · 152 tokens out

• Claude Sonnet 5.5

  Failed

  Graded 4.0 of 5 on average (lowest 3); but 181 words, over the 150 allowed.

  **Part-Time Bookkeeper – Leeds Architecture Studio**
  
  We're a friendly 12-person architecture studio in Leeds looking for a part-time bookkeeper to keep our finances accurate and organised.
  
  **What you'll do**
  - Process sales and purchase invoices and reconcile bank accounts
  - Run monthly payroll and manage expenses
  - Prepare VAT returns and monthly management reports
  - Chase late payments and liaise with our accountant
  
  **What we need**
  - Proven bookkeeping experience, ideally in a small business
  - Confident with cloud accounting software (Xero preferred)
  - Careful, organised and comfortable working independently
  - AAT or equivalent qualification is helpful, but not essential
  
  **What we offer**
  - £16,000 a year (20 hours a week, pro rata from a full-time salary of £32,000)
  - Flexible days, with a mix of studio and home working
  - A supportive team in a creative, welcoming workplace
  - 28 days' holiday, pro rata
  
  To apply, send your CV and a short cover note to [email address].
  
  ---
  
  Note: I assumed the £16,000 is the actual earnings figure for 20 hours, so I've stated it that way. I also invented the holiday, flexibility and software details, so please adjust them to match your studio.

  1 message on Pro · $0.0054 to run · 4.1 s · 426 tokens out

• Gemini 3.8 Flash

  Failed

  Graded 3.3 of 5 on average (lowest 2).

  **Part-Time Bookkeeper | Leeds**  
  **Salary:** £16,000 per annum (20 hours/week)
  
  We are a collaborative, 12-person architecture practice based in central Leeds looking for an organised Bookkeeper to manage our day-to-day finances.
  
  **What you’ll do:**
  * Manage sales and purchase ledgers, including invoicing and supplier payments
  * Perform bank and credit card reconciliations
  * Track project expenses and assist with monthly financial reporting
  * Prepare quarterly VAT returns and liaise with our external accountant
  
  **What we need:**
  * Proven bookkeeping experience (AAT qualified or qualified by experience)
  * Proficiency with cloud accounting software (Xero preferred) and Excel
  * Strong attention to detail and good communication skills
  
  **What we offer:**
  * Flexible working hours (spread across 3–4 days to suit you)
  * Friendly, design-led studio environment
  * Hybrid working options after onboarding
  * 25 days annual leave (pro-rata) plus bank holidays
  
  To apply, please send your CV to [Email Address].

  1 message on Pro · $0.0006 to run · 1.8 s · 230 tokens out

Giving the model the alternative you can afford keeps it from inventing a smaller discount you never agreed to.

Write an email to a long-standing customer who asked for a 20% discount. Say no, offer free delivery for the next six months instead, and keep them. Under 120 words.

• GPT-6 Luna

  Passed

  Graded 5.0 of 5 on average (lowest 5).

  **Subject: A thank-you for your loyalty**
  
  Hi [Name],
  
  Thanks for being a valued customer for so many years. I understand you’re looking for a 20% discount, but we’re unable to offer that.
  
  As a thank-you for your continued business, we’d be happy to provide free delivery on your orders for the next six months. We truly appreciate your loyalty and hope this helps make it easier to keep working together.
  
  Best,  
  [Your Name]

  1 message on Pro · under $0.0001 to run · 2.0 s · 99 tokens out

• GPT-6 Sol

  Passed

  Graded 5.0 of 5 on average (lowest 5).

  **Subject: Thank you for your continued partnership**
  
  Hi [Name],
  
  Thank you for being a customer for so many years. We value your business and appreciate you asking about a 20% discount.
  
  We can’t offer that discount, but we’d like to provide **free delivery on your orders for the next six months**. We hope that helps reduce your costs, and we’d welcome the chance to keep working together.
  
  Please let me know if you’d like us to apply free delivery to your account.
  
  Best,  
  [Your Name]

  1 message on Pro · $0.0009 to run · 1.9 s · 113 tokens out

• Claude Sonnet 5.5

  Passed

  Graded 4.7 of 5 on average (lowest 4).

  **Subject:** Your request, and something we can do instead
  
  Dear [Name],
  
  Thank you for your message, and for the years you've trusted us with your business. It genuinely means a lot to our team.
  
  I've given your request for a 20% discount careful thought. Unfortunately, we're unable to offer that level of reduction without compromising the quality and service you've come to expect from us.
  
  What I can do is offer you **free delivery on all orders for the next six months**, effective immediately. I hope it shows how much we value having you as a customer.
  
  If you'd like to talk it through, I'm happy to call at your convenience.
  
  Warm regards,
  [Your Name]
  [Company]

  1 message on Pro · $0.0031 to run · 2.5 s · 211 tokens out

• Gemini 3.8 Flash

  Passed

  Graded 4.7 of 5 on average (lowest 4).

  Subject: Update on your recent discount request
  
  Hi [Customer Name],
  
  Thank you for reaching out, and more importantly, thank you for your continued partnership over the years. 
  
  While we can’t accommodate a 20% discount on our pricing, we truly value your loyalty and want to show our appreciation. As a token of our gratitude, we would love to offer you complimentary delivery on all your orders for the next six months. 
  
  We hope this helps reduce your overall costs while allowing us to maintain the high quality of service you expect from us. 
  
  Thank you again for being a valued part of our community.
  
  Best regards,
  
  [Your Name]  
  [Your Company]

  1 message on Pro · $0.0004 to run · 1.8 s · 144 tokens out