Issue 28: Smarter by Thursday
Smarter by Thursday
One practical AI win, every week. No jargon required.
We’re doing two very practical things this week: **using AI to research a major purchase** and **using AI to write a professional “no” email when you don’t want to burn a bridge**. These both hit the same nerve: decision fatigue. Whether you’re staring at 14 nearly identical dishwashers or agonizing over how to decline a request without sounding cold, your brain is getting tired in the same way.
Try at least one before Sunday. That is the whole assignment.
Use Case 1 of 2
Use Case 1: Researching a major purchase decision
On a Tuesday evening in July, around 8:30 p.m., David is sitting at his kitchen table with his laptop open and three browser tabs fighting for his attention. He’s 56, his kids have mostly launched, and the family minivan just died in spectacular fashion in the grocery store parking lot.
He’s been “just checking options” for four nights in a row. Now he has: - a dealer’s quote in his email, - a YouTube video insisting he should only buy used, - and a spreadsheet his brother-in-law sent with interest rates and lease vs. buy numbers.
He’s not clueless. He understands budgets. He can read reviews. But he’s drowning in *almost* useful information. Every time he narrows it down, he sees one more review about a transmission issue or a horror story about the dealership, and he’s back to square one.
What David really wants is a calm, neutral second brain that: - summarizes the tradeoffs, - surfaces *real* concerns (not Internet drama), - and helps him decide what to research next instead of opening ten new tabs.
That is exactly where AI is useful. Not to decide *for* him, but to organize the mess and point out what actually matters for *his* situation and budget.
Here is the exact prompt to use:
Copy and paste this into ChatGPT or Claude:
Copy and paste this, then fill in the brackets with your details:
I’m planning a major purchase and I want you to act as a calm, neutral research assistant.
Here is my situation: - What I’m buying: [e.g., midsize hybrid SUV, new or 2 - 3 years used] - My budget range: [total price range or monthly payment range] - How long I plan to keep it: [e.g., at least 8 - 10 years] - My main priorities (ranked): [e.g., reliability, safety, fuel costs, comfort, resale value] - My constraints: [e.g., must fit in small garage; prefer local dealer; avoid complicated tech] - My rough shortlist so far: [list 2 - 5 models or options, or say “I don’t have a shortlist yet”]
Step 1: Based on my situation, list the 5 - 7 most important decision factors I should care about for this purchase (in plain language).
Step 2: For each option on my shortlist, give me a simple pros/cons table based on: - reliability and known issues - total cost of ownership over the time I’ll own it (not just sticker price) - practicality for my specific needs - any “red flag” concerns I should double-check with my own research
Step 3: Suggest 5 - 7 concrete things I should do next (specific sites to check, numbers to ask the dealer for, types of reviews to read) *without* assuming I’m tech-savvy.
Step 4: Ask me 5 clarifying questions that would let you refine the advice further for my situation.
Keep the tone calm and matter-of-fact. Avoid hype. If there are uncertainties or tradeoffs, spell them out directly instead of glossing over them.
Why this prompt works:
You’re not just saying “help me pick a car.” You’re **giving context** (budget, time horizon, priorities), and then **breaking the AI’s job into steps**: define decision factors, compare options, suggest next actions, and ask clarifying questions. That structure steers the model away from vague “Top 10 SUVs” fluff and toward a customized, side‑by‑side view that reflects what *you* care about. It also explicitly invites the AI to admit uncertainty, which is where these tools tend to mislead people.
AI will speak confidently about things it does not actually “know” about your local market - like real-time prices or dealer behavior. Treat any numbers it gives as **ballpark**, not gospel. Always cross-check prices, reliability information, and “known issues” with at least one independent source (consumer sites, manufacturer pages, or trusted review outlets) before spending real money. And don’t let a slick AI summary override your own gut if something feels off; the model cannot see the car, the dealer, or your bank account.
Use Case 2 of 2
Use Case 2: Writing a professional decline email
Thursday afternoon, 4:15 p.m., inbox open, coffee cold.
Sandra is 49 and runs a small consulting practice she built after leaving a corporate role. A former colleague has just emailed asking if she could “jump on a quick project” for a fraction of her usual rate - “it’ll be great visibility.”
She does not want to say yes. She also does not want to sound ungrateful, unhelpful, or “too big for her britches.” So she does what many of us do: - types a long, apologetic draft, - deletes it, - types a short, blunt draft, - worries it sounds rude, - leaves the email sitting there for three days.
Meanwhile, the mental load is real. Her brain is quietly chewing on “What if he thinks I’m not a team player?” while she’s trying to focus on paying clients.
What she needs is a way to: - say **no** clearly, - acknowledge the relationship, - offer an alternative *only if she genuinely wants to*, - and do it in 5 minutes instead of spinning for an hour.
AI can’t decide your boundaries, but it can give you a clean first draft that sounds like a grounded adult instead of an anxious over-explainer.
Here is the exact prompt to use:
Copy and paste this into ChatGPT or Claude:
Copy and paste this, then fill in the brackets with your details. Do **not** paste the other person’s personal details - summarize them instead.
I need help writing a professional email where I decline a request without damaging the relationship.
Context (please read this carefully): - Who they are: [e.g., former colleague, good relationship but not close friend] - What they asked for: [brief description, no names or sensitive details] - Why I want to say no: [e.g., too low budget, misaligned with my services, over capacity] - How I usually communicate with them: [formal, semi-formal, casual but respectful] - What I’m okay offering instead (if anything): [e.g., referral to someone else, a brief call in 2 weeks, or “nothing - just a clear no”]
Write an email I can send that: - clearly declines the request in one or two sentences - briefly but honestly explains my reason at a high level (no oversharing) - expresses appreciation for them thinking of me - keeps the door open in a realistic way (only if that fits the context I gave you) - matches the communication style I described - is 150 words or less
After the draft, list 3 short variations of the key “decline” sentence so I can choose the one that feels most natural.
Do not invent details I didn’t give you. Do not apologize excessively - one brief apology is enough.
Why this prompt works:
You’re not saying “write a nice no email.” You’re **laying out the social context** (who they are, your relationship), **explaining your reasons**, and **setting constraints** (length, tone, no over-apologizing). That gives the AI enough structure to stay out of the two ditches people hate: stiff corporate jargon on one side, and overly gushy, self-sacrificing apologetics on the other. Asking for a few alternate decline sentences lets you tune the language to what feels like *you.*
AI will sometimes overdo either the politeness (“profusely apologizing”) or the sales pitch (“I’d love to explore other ways to partner!” when you absolutely would not). Read the draft out loud before sending. If any line makes you roll your eyes or feels out of character, edit it or delete it. And never paste confidential or sensitive details about the other person or the project; summarize instead.