3 ChatGPT Prompts to Make Money Fast

Let me be straight with you before you copy a single prompt: ChatGPT will not make you money. People with money and problems will make you money. What ChatGPT can do — and does remarkably well when you prompt it properly — is compress the three slowest, most procrastination-prone steps between "I should earn some extra income" and "someone just paid me."

Those three steps are:

1. Figuring out what you can actually sell. Most people are terrible at seeing their own skills. You've been doing your job so long that the hard parts feel obvious, and the obvious parts feel worthless. They're not.
2. Turning a vague skill into a concrete offer. "I'm good with spreadsheets" earns nothing. "I'll clean up your bookkeeping backlog in one weekend for $400" earns money. The distance between those two sentences is where 90% of would-be side incomes die.
3. Actually telling people. Not strangers on the internet. The 50–200 people who already know you, trust you, and would happily pay you or refer you — if only they knew you were offering something.

This post gives you one deep prompt for each step. Not "act as a millionaire mentor" fluff — real working prompts with structure, constraints, and output formats, the kind I actually use. Each one is fully written out so you can copy and paste it, and then I'll dissect why it's built the way it is, walk through a fictional worked example, show you how to customize it, and explain how to fold it into a weekly routine so this becomes a repeatable workflow instead of a one-time burst of motivation.

A note on the word "fast," because I refuse to be another guy promising you $10k by Friday. In this context, fast means: a real offer, a real pitch, and ten real conversations started within seven days — and, if your network is warm and your skill is in demand, your first paid work within two to four weeks. That is genuinely fast for service income. Anyone promising faster is selling you a course. I go deeper on realistic timelines at the end of this post, and if you want the full menu of income models beyond services, I wrote up twelve of them in 12 Real Ways to Make Money With AI in 2026.

One more thing: these prompts work in ChatGPT, Claude, Gemini — any capable model. I test everything in at least two. The structure is what matters, not the logo on the chat window.

Let's go.

---

Prompt 1: The Skills Audit

What it does: interviews you about your work history, extracts skills you didn't know were sellable, and outputs five concrete offers ranked by speed-to-cash.

The reason to start here — rather than brainstorming offers directly — is that you are the worst-positioned person in the world to evaluate your own skills. The curse of knowledge means everything you're good at feels easy, and things that feel easy feel unsellable. An HR manager thinks "anyone can write a job description." A bookkeeper thinks "anyone can reconcile accounts." Meanwhile, small business owners are paying real money every week for exactly those things, badly done.

This prompt forces the model to act as an outside observer with a commercial lens. Copy everything in the block below into a fresh chat:

You are a pragmatic career strategist who helps professionals turn their
existing skills into paid side income. You are allergic to hype. You never
suggest "start a faceless YouTube channel" or "do dropshipping." You only
recommend offers where the person's EXISTING experience is the product.

Your job has two phases. Do not skip Phase 1.

PHASE 1 — INTERVIEW ME.
Ask me the following questions ONE AT A TIME. Wait for my answer before
asking the next. Push back if my answers are vague — ask a follow-up until
you get specifics.

1. What is your current job title, and what were your last two before it?
2. Describe a normal Tuesday at work. What do you actually DO, hour by hour?
3. What do colleagues come to you for help with, even when it's not your job?
4. What task do you finish in 1 hour that takes others a full day?
5. What have you done at work that saved money, made money, or prevented
   a disaster? Give numbers if you can.
6. What tools/software do you know well enough to teach?
7. Who do you know? Describe your network: industries, seniority, roughly
   how many people would recognize your name warmly.
8. How many hours per week can you realistically sell (be honest), and
   what is the minimum $/month that would make this worth doing?
9. What will you absolutely NOT do? (e.g., no phone calls, no weekends,
   no work for your current employer's competitors.)

PHASE 2 — BUILD MY OFFER SHORTLIST.
Once you have my answers, produce exactly this output:

A) SKILLS INVENTORY — a table with 3 columns:
   [Skill] [Evidence from my answers] [Who pays for this and why].
   Include at least 8 skills. At least 3 must be skills I did NOT
   explicitly name — infer them from my stories.

B) FIVE SELLABLE OFFERS — for each offer give:
   - Name of the offer (specific, e.g. "New-manager onboarding kit for
     10–50 person companies", not "HR consulting")
   - The exact buyer (job title + company size + situation they're in)
   - The painful problem it solves, in the buyer's own words
   - Deliverable and rough scope (what they get, how long it takes me)
   - Price range for a first client (conservative, real-world)
   - Speed-to-cash score, 1–10, where 10 = "could be paid within 14 days
     via my warm network" — and one sentence justifying the score
   - The single biggest risk that this offer flops

C) RANKING — rank the 5 offers by speed-to-cash, and tell me which ONE
   you would pursue first given my hours and income target from Q8, and
   why. Be decisive. Do not hedge with "it depends."

Rules for the whole exercise:
- Ground every offer in evidence from MY answers. If you can't tie an
  offer to something I said, don't suggest it.
- Assume I keep my day job. Every offer must fit the hours from Q8.
- Prefer offers sold to businesses over consumers (they pay faster and
  more), unless my network is clearly consumer-facing.
- No offers that require an audience, a website, or paid ads to start.
- Flag any offer with a conflict-of-interest risk against my employer.

Begin Phase 1 now with question 1.

Why this prompt works

It interviews you one question at a time. If you ask ChatGPT "what can I sell?" in one shot, you'll type two lazy sentences about yourself and get generic advice back — garbage in, garbage out. The one-at-a-time interview format forces you to produce the raw material the model needs, and the instruction to "push back if my answers are vague" makes the model do what a good coach does: refuse your first, boring answer.

Question 3 and question 4 are the money questions. "What do colleagues come to you for?" and "what takes you an hour that takes others a day?" surface the skills you've stopped noticing. In my experience these two questions produce the winning offer more often than everything else combined.

The evidence rule kills hallucinated advice. "Ground every offer in evidence from MY answers" is the single most important line. Without it, the model drifts into suggesting whatever's trendy. With it, every suggestion must trace back to something you actually said.

Speed-to-cash scoring forces honest ranking. Notice the definition baked in: 10 means paid within 14 days via your warm network. That definition matters. An offer can be lucrative but slow (corporate training — great money, three-month sales cycles) or modest but fast (résumé rewrites — paid this week). When your goal is momentum, you want fast first. You can upsell later.

The constraints section protects you from fantasy. No audience required, no ads, no quitting your job, fits your real hours. This is the difference between advice for your actual life and advice for a hypothetical person with infinite time.

Worked example: Dana, HR manager

Dana is a fictional-but-realistic HR manager at a 400-person logistics company. Eleven years in HR, the last four as a manager. She runs the prompt. Her interview answers reveal: she's rewritten the company's entire interview process, colleagues constantly ask her to "look over" offer letters and tricky termination emails, she can produce a compliant job description in 20 minutes, and her network is full of ops managers and small-agency owners from a previous job.

The skills inventory surfaces three things Dana never would have listed: "de-escalation writing" (from her stories about termination emails), "interview loop design," and "compensation benchmarking for non-tech roles." Her five offers come back roughly like this:

1. Hiring-process-in-a-box for 10–50 person companies — interview scorecards, question banks, and a structured loop, delivered in two weeks. $1,200–$2,500. Speed-to-cash: 8.
2. Job description + offer letter package — five roles rewritten for clarity and compliance. $400–$700. Speed-to-cash: 9.
3. "Difficult conversation" script service for founders — terminations, demotions, PIP conversations. $250 per script. Speed-to-cash: 7 (buyers only need it when it hurts).
4. New-manager onboarding workshop — half-day, delivered over Zoom. $900–$1,500. Speed-to-cash: 5 (scheduling drags).
5. Fractional HR retainer — 5 hrs/month for companies too small for an HR hire. $500–$900/month. Speed-to-cash: 4 (great later, slow to close first).

The model recommends offer #2 as the wedge — small, cheap, fast, and a natural upsell into #1 and #5 — and flags that offer #5 could conflict with her employment contract. That's a genuinely useful strategic read, and Dana got it in 25 minutes.

Customization tips

• If you're deep in one industry, add a line: "My buyers will almost certainly be in [industry]. Weight offers toward problems specific to it." Niche offers close faster because the buyer thinks "this person gets my world."
• If you have zero evening energy, change Q8's framing to "hours I can sell before work or on Saturday morning" — the offers will shift toward async deliverables instead of calls.
• If the five offers feel generic, reply with: "These are too generic. Regenerate offers 2–5, each at least twice as narrow. Narrow the buyer, not just the deliverable." Narrowness is a dial you can keep turning.
• If you want a sanity check, paste the final shortlist into a different model (Claude, Gemini) with: "Red-team these five offers. Which would you refuse to buy, and why?" Cross-model criticism is cheap due diligence. I use the same trick in 6 Claude Prompts That Put You Ahead of 99% of Users.

Making it a routine

Run the full audit once. Then put a 20-minute block in your calendar every Sunday: paste your shortlist back into the chat and add "Here's what happened this week: [what you tried, who replied, what they said]. Update the speed-to-cash scores and tell me if the evidence supports switching offers." The audit stops being a one-off quiz and becomes a living document that sharpens with real market feedback. That feedback loop — not the initial output — is where the compounding happens.

---

Prompt 2: The Offer Builder

What it does: takes the one offer you chose and turns it into a one-paragraph pitch, a three-tier price structure, and a 7-day launch checklist.

An offer that lives in a chat log earns nothing. The gap between "I know what I could sell" and "I have words I can say to a human and a price I can defend" is where the second wave of people quit. This prompt closes that gap in one sitting.

Start a new chat (fresh context keeps the model focused) and paste this, filling in the bracketed section at the bottom:

You are a no-nonsense offer strategist for solo professionals starting a
service side business. You have two obsessions: (1) pitches written in
the buyer's language, not the seller's, and (2) getting the first paying
client within 14 days. You hate jargon, "passion," and mission statements.

I will give you ONE offer I've chosen to pursue. Turn it into a complete
launch package with exactly these four sections:

SECTION 1 — THE PITCH PARAGRAPH.
Write a 4–6 sentence pitch I can say out loud or paste into a message.
Structure: (a) the specific problem as the buyer experiences it,
(b) what I deliver and how fast, (c) one line of credibility drawn from
my background, (d) a low-friction next step (a 20-minute call, not
"buy now"). Then give me TWO alternates: one more formal, one more
casual. Rule: a smart 14-year-old should understand every sentence.
Banned words: solutions, leverage, passionate, journey, unlock, seamless,
holistic, empower, synergy.

SECTION 2 — PRICING.
Give me three tiers: Starter / Standard / Premium.
For each tier: name, exact price (a number, not a range), what's
included, delivery time, and who should pick it.
Rules:
- Anchor prices to the VALUE of the outcome, and sanity-check them
  against what freelancers/agencies actually charge for similar work.
- Starter must be an easy yes for a warm contact: priced so a manager
  could expense it without approval, scoped to deliver in ≤1 week.
- Premium should be 4–6x Starter and include the most hand-holding.
- Then tell me: which tier to lead with for my FIRST client and why,
  and give me one sentence to say out loud when someone asks
  "how much?" — including how to stay silent after saying the number.
- Also give a "first three clients" founding discount policy: how much,
  what I ask for in return (testimonial + referral), exact wording.

SECTION 3 — 7-DAY LAUNCH CHECKLIST.
A day-by-day plan, Days 1–7, max 90 minutes per day. Each day: the
concrete tasks, what "done" looks like, and the single metric to hit.
Constraints:
- No website. No logo. No business name debates. No paid ads.
- Day 1–2 = assets (pitch finalized, one-page offer doc or PDF,
  payment link set up).
- Day 3–5 = outreach to warm contacts (I have a separate system for
  writing the messages; just tell me the volume and cadence).
- Day 6–7 = follow-ups, first calls, and adjusting the pitch based
  on real replies.
- Include exactly ONE "kill criterion": a specific signal by Day 7
  that means I should change the offer or the audience before
  pushing harder. Be precise about the threshold.

SECTION 4 — OBJECTIONS.
The five most likely objections I'll hear (e.g., "can't you just do
this with ChatGPT?", "we don't have budget", "send me some info"),
each with a two-sentence response that is honest, not slick, and a
note on when the correct response is to walk away.

Here is my offer:
[PASTE: the offer from your Skills Audit — the offer name, the buyer,
the problem, the deliverable, the price range, plus 2–3 sentences about
your background and anything relevant about your network.]

Why this prompt works

The pitch structure is problem-first. Almost everyone pitches backwards: "I'm an HR professional with 11 years of experience offering..." Nobody cares. The structure forces the buyer's problem into the first sentence, which is the only sentence that determines whether they keep reading. The banned-words list matters more than it looks — those words are how pitches die. When the model can't reach for "leverage" and "seamless," it's forced to write like a person.

Three tiers with exact numbers beats "what should I charge?" Open-ended pricing questions get you mush ("it depends on your market!"). Demanding exact numbers with structural rules — Starter expensable without approval, Premium at 4–6x — gets you a defensible structure you can adjust, which is psychologically completely different from a blank page. And the "say the number, then stay silent" coaching addresses the actual failure mode: new sellers quote a price and immediately discount it themselves out of nervousness, before the buyer has said a word.

The 90-minute daily cap makes the checklist survive contact with your real life. Launch plans fail because Day 1 says "set up your brand" and eats a weekend. Capping each day and banning websites, logos, and business-name debates strips out every form of productive procrastination. You are left with only the tasks that lead to a paying client.

The kill criterion is the most honest line in the prompt. Most money-with-AI content pretends every offer works. Real ones sometimes don't. Defining in advance what failure looks like ("if I've reached 10 warm contacts and gotten zero calls booked, the offer or audience is wrong") turns a demoralizing week into a data point, and stops you from either quitting too early or grinding a dead offer too long.

Worked example: Marcus, high school teacher

Marcus (fictional) teaches high school English and runs the school's debate team. His Skills Audit surfaced an offer he'd never have called an offer: parents in his community regularly ask him for help with their teenagers' college application essays, and he's naturally good at coaching kids to sound like themselves instead of like a brochure.

The Offer Builder gives him a pitch paragraph that leads with the parent's actual fear: "Every fall I watch smart kids write college essays that sound like they were written by a committee — and admissions readers can tell. I coach students through three drafts in three weeks until the essay sounds unmistakably like them. I've spent twelve years teaching teenagers to write, and I run our school's debate program. If your teen is starting applications, I'll do a free 20-minute read of their current draft and tell you honestly whether they need help."

Pricing comes back as: Starter — single essay review with recorded video feedback, $95, delivered in 3 days. Standard — three-draft coaching package with two calls, $350. Premium — full application season support, up to five essays, $1,400. The model tells him to lead with Starter in September (parents are panicking, $95 is an impulse yes) and notes his real business is the upsell from Starter to Standard. His kill criterion: if 10 parents see the offer and none books the free 20-minute read, the problem is trust or timing, not price — so shift from messaging parents directly to asking two fellow teachers for referrals.

The objections section correctly predicts "can't my kid just use ChatGPT to write it?" and hands him the honest answer: yes, and admissions officers are now very good at noticing, which is exactly why coaching a student to write authentically is worth more than it was five years ago.

Total time: about 40 minutes, including his edits. Marcus has a pitch, prices, and a plan. Note the ethical guardrail the prompt's honesty rules enforce: he's coaching essays, not writing them — an offer that survives scrutiny.

Customization tips

• Paste in real market data if you have it. If you know what local competitors charge, add it: "For reference, [competitor type] charges $X for [thing]." The pricing output improves dramatically with even one real anchor.
• If the pitch sounds like ads copy, reply: "Rewrite this the way I'd explain it to a former colleague over coffee. Shorter sentences. No enthusiasm." The casual alternate is usually the one that works in warm outreach anyway.
• If you're productizing rather than doing custom work — fixed scope, fixed price, repeatable delivery — say so explicitly; the checklist changes shape. I'm biased toward productized offers because they're the most repeatable; see the productized AI service skill for a full template, and How to Build a One-Person Business With AI in 30 Days for the longer arc this 7-day sprint plugs into.
• Stress-test the price. Follow up with: "Argue that my Standard tier is priced 40% too low. Now argue it's 40% too high. Which argument is stronger?" You'll usually discover you're underpricing — new sellers almost always are.

Making it a routine

The launch checklist is a one-week sprint, but the Offer Builder itself becomes your iteration engine. Every Sunday, in the same 20-minute block from Prompt 1, paste in what actually happened: which pitch variant you used, what people said on calls, which objections came up that the model didn't predict. Ask it to revise the pitch and pricing accordingly. After three or four cycles, your pitch will be written mostly in your buyers' own recycled words — which is exactly what a good pitch is.

---

Prompt 3: The Outreach Engine

What it does: turns your contact list into ten personalized, non-cringe messages to warm contacts, plus a follow-up sequence and a tracking system.

This is the prompt people avoid, because it's the one that involves other humans. But it's also where the money is. Your first clients will almost certainly come from people who already know you — not because of some networking secret, but because trust is the slowest thing to build and you've already built it. A warm contact who ignores your message costs you nothing. A warm contact who replies "oh interesting, actually my sister needs exactly this" is how side incomes actually start.

The failure mode is sending ten copies of the same stiff announcement. This prompt prevents that. New chat, paste, fill in the brackets:

You are an outreach writer for a professional launching a service side
business to their warm network. Your style: warm, brief, specific, zero
sales-speak. Every message must sound like it was written by a busy
human to a person they genuinely know — because it was. If a message
could be sent to two different people unchanged, you have failed.

CONTEXT
My offer (pitch paragraph): [PASTE YOUR PITCH FROM THE OFFER BUILDER]
About me: [2–3 sentences: job, background, why I'm starting this now]

MY 10 WARM CONTACTS
For each I'll give: name / how I know them / last real interaction /
what they're up to now / why I think they might care (buyer themselves,
knows buyers, or just supportive and well-connected).
1. [ ... ]
2. [ ... ]
(through 10)

YOUR TASKS

TASK 1 — SORT. Classify each contact: BUYER (might pay me directly),
CONNECTOR (likely knows buyers), or SUPPORTER (cheerleader, low
referral odds). Tell me the ideal order to contact them and why —
I want practice reps and quick feedback before the highest-stakes
messages.

TASK 2 — WRITE 10 FIRST MESSAGES. For each contact, in their channel
of choice (assume text/WhatsApp for close friends, LinkedIn DM or
email for professional ties — infer from how I know them). Rules:
- Under 120 words. Close friends: under 60.
- Open with something real about THEM from my notes — not
  "hope you're well." If my notes are too thin to do this honestly,
  tell me to go check their LinkedIn first rather than faking it.
- One line on what I'm now offering, in plain words.
- The ask must match their category: BUYERS get offered the
  low-friction next step from my pitch. CONNECTORS get "does anyone
  come to mind?" — never a hard sell. SUPPORTERS get a simple
  "keep me in mind."
- Include exactly ONE easy-to-answer question per message.
- Never apologize for reaching out. Never say "shameless plug,"
  "sorry to bother you," or "no worries if not!!"
- It must be OBVIOUS I'm now doing this for money. No bait-and-switch
  coffee chats that turn into a pitch.

TASK 3 — FOLLOW-UPS. For each category, write:
- Follow-up 1 (day 4–5 after silence): adds one NEW piece of value
  or context — a relevant thought, a useful link, a sharper version
  of the offer. Never "just bumping this."
- Follow-up 2 (day 10–12): a graceful close that leaves the door
  open, under 40 words.
- Hard rule: two follow-ups maximum, then stop. Write the rule into
  the plan so I can't rationalize a third.

TASK 4 — REPLY PLAYBOOK. Short scripts for the five replies I'll
actually get: (1) "Tell me more," (2) "How much?", (3) "Not for me,
but try [name]," (4) "Wow, congrats!" with no follow-through,
(5) silence. For #3, include how I ask for a warm intro instead of
a cold name-drop.

TASK 5 — TRACKER. A simple table I can keep in a note or
spreadsheet: contact, category, sent date, follow-up dates, status,
next action. Fill in the first version based on your contact order
with real dates starting tomorrow, spread over 5 days — never all
ten in one day, so early replies can improve later messages.

Why this prompt works

The sorting step changes the psychology. Treating all ten contacts identically produces either ten timid messages or ten pushy ones. Once contacts are sorted into buyers, connectors, and supporters, each message gets an ask that actually fits the relationship — and connectors, counterintuitively, are usually worth more than buyers. One well-connected former colleague who mentions you in the right meeting outperforms five direct pitches.

"If a message could be sent to two different people unchanged, you have failed" is the line doing the heaviest lifting. It converts personalization from a nice-to-have into a pass/fail condition the model checks itself against. And the companion rule — if the notes are too thin, tell me to go look, don't fake it — keeps the model from inventing fake familiarity ("loved your recent post!") that would torch real trust.

The honesty rules protect your network, which is worth more than any single client. No fake coffee chats, no burying the commercial intent, a hard cap of two follow-ups. Warm outreach goes wrong when people feel used. Every rule in this prompt exists so that the worst case is a contact thinking "good for them, not for me" — a perfectly fine outcome that leaves the relationship intact for next time.

Staggered sending builds in a feedback loop. Sending all ten messages at once means all ten carry the same mistakes. Spreading them over five days, ordered from low-stakes to high-stakes, means your fifth message benefits from what the first four taught you. This is the difference between a broadcast and a learning system.

The reply playbook removes the freeze. The scariest moment isn't sending — it's when someone replies "how much?" and you have 90 seconds to answer without panicking and discounting. Having the scripts ready, including the graceful ask for a warm intro, means you respond from preparation rather than adrenaline.

Worked example: Priya, bookkeeper

Priya (fictional) has done bookkeeping at a mid-size firm for eight years. Her chosen offer, from Prompts 1 and 2: a fixed-price "books cleanup" for small business owners who are 3–12 months behind — $500 flat for a diagnostic and cleanup plan, delivered in one week, with monthly bookkeeping as the natural follow-on.

She lists ten contacts: her cousin who owns a landscaping company, two former colleagues now at other firms, a friend who runs a yoga studio, her old accounting professor, a neighbor who's a realtor, and four assorted friends in small-business-adjacent jobs. The model sorts them: cousin and yoga-studio friend are BUYERS (both plausibly behind on their books), the realtor and professor are CONNECTORS (realtors know every small business owner in town; professors know alumni), the rest split between connector and supporter.

Her message to the yoga-studio friend comes out at 54 words, opens with a reference to the studio's recent second location — which is exactly why her books are probably a mess — states the offer in one plain sentence, and asks a single question: "Want me to take a look before your tax deadline sneaks up?" Her message to the realtor doesn't pitch at all; it asks, "You know basically every business owner in town — does anyone come to mind who's behind on their books and stressed about it?"

The tracker schedules the two easiest supporter messages for Monday, buyers midweek, and the high-value connectors Friday, after she's warmed up. By the following Wednesday, in this fictional-but-typical scenario: four replies, one "congrats" with nothing behind it, one referral from the realtor, and her cousin — predictably — becomes client number one at a family discount in exchange for a testimonial. That's not wealth. It's ignition, and ignition is the hard part.

Customization tips

• Can't think of ten people? You're defining "warm" too narrowly. Tell the model: "Help me build the list first — walk me through categories: former colleagues, former managers, clients from past jobs, classmates, neighbors, people from my hobby, other parents at school." Most people find thirty names in fifteen minutes.
• Feed real replies back in. When someone responds, paste the exchange into the chat: "Here's what she said. Draft my reply — same tone, keep it short." The model is even better at continuing conversations than starting them.
• Match your actual voice. Paste in two or three real messages you've sent to friends and add: "Match this voice exactly, including how casual my punctuation is." A message that's 20% more polished than you actually are reads as fake to people who know you.
• After the warm ten, the same structure extends to second-degree outreach (people your connectors introduce you to) — just tell the model the intro context and it will calibrate the added formality. Don't skip to cold outreach until warm is exhausted; the economics are wildly different.

Making it a routine

Outreach is not a launch event; it's a habit with a small weekly quota. Mine looks like this, and I recommend copying it: every week, five new conversations and all due follow-ups, tracked in the table, 45 minutes total. Each Sunday, in the same session as your offer review, paste your tracker into the chat: "Here's the current state. Write this week's five new messages and the follow-ups that are due." The model handles the drafting; you handle the judgment and the send button. Ten warm contacts is the first tank of fuel — the routine is the engine.

---

What "fast" really means: the honest caveats

I promised zero hype, so here is the fine print most posts like this omit.

"Fast" means fast for service income, not fast in absolute terms. Realistic timeline if you execute all three prompts with a warm network and an in-demand skill: offer defined in week one, ten conversations started by week two, first paid project somewhere in weeks two to six. First month's income: usually a few hundred to a couple thousand dollars, not five figures. The people who show screenshots of more either had an audience already, got lucky, or are lying to sell you something.

The prompts compress thinking, not trust. ChatGPT can produce your offer in an hour, but it cannot make your cousin trust you — eight years of being reliable did that. If your network is thin or cold, these prompts still work; the timeline just stretches, because you'll be building trust and pitching at the same time. That's fine. Start anyway. The network you have in six months is built from the conversations you start this week.

Speed and durability trade off. The fastest offers (small fixed-scope services to warm contacts) are rarely the biggest ones. That's the correct trade at the start: your first goal isn't maximizing income, it's completing one full loop — offer, pitch, client, delivery, testimonial — because the loop is what you'll repeat and scale. Durable models come after the loop works; that's the subject of 12 Real Ways to Make Money With AI in 2026 and 12 Ways to Make Money With Claude.

Delivery is still on you. Nothing in this post helps you if the work you deliver is bad. Only sell what you can genuinely do well, use AI to speed up your delivery too (that's where skills like meeting-notes-to-actions and a proper AI marketing setup come in), and over-deliver shamelessly for your first three clients. Their testimonials and referrals are worth more than their fees.

Check your employment contract. Moonlighting clauses, non-competes, IP assignment. Boring, and much cheaper to check now than later. Prompt 1 flags conflicts, but a model is not a lawyer.

Expect a mostly quiet week one. Six of your ten contacts will not reply, and that is a normal, healthy result — not a verdict on you. The system works on rates, not on any single message. That's precisely why it's built as a weekly routine instead of a one-time blast: repetition is the entire trick. If you want to get better at the prompting side of this — why these prompts are structured the way they are, and how to build your own — the full breakdown is in The Complete Guide to Prompt Engineering in 2026.

The whole system on one page

• Sunday, 45–60 minutes: run or update the Skills Audit review, revise the pitch with the week's real feedback, generate the week's five outreach messages and due follow-ups.
• Monday–Friday, ~10 minutes a day: send what's scheduled, log replies in the tracker, paste any live conversations into ChatGPT for reply drafts.
• Every day 7: check the kill criterion. Pass — keep going. Fail — change the offer or the audience, not the effort level.
• After your first paid client: deliver exceptionally, ask for a testimonial and one referral, raise your Starter price 20%, and keep the routine running.

Three prompts, one hour a week, and the skills you already have. That's the whole system. It's not magic, and that's exactly why it works.

Resources & links

• 12 Ways to Make Money With Claude — the Claude-specific playbook of income models, from services to skills.
• 12 Real Ways to Make Money With AI in 2026 — the full menu of durable AI income models with realistic earning ranges.
• How to Build a One-Person Business With AI in 30 Days — the 30-day arc your 7-day launch sprint plugs into.
• 6 Claude Prompts That Put You Ahead of 99% of Users — advanced prompting patterns, including the red-teaming trick from this post.
• The Complete Guide to Prompt Engineering in 2026 — why these mega-prompts are structured the way they are, and how to design your own.
• Productized AI Service skill — a ready-made template for turning your offer into a fixed-scope, fixed-price service.
• Cover Letter Writer skill — the same "buyer's language, zero fluff" writing method, applied to landing your next role.
• AI Marketing Team skill — build out your marketing once the warm-network phase is exhausted.
• Meeting Notes to Actions skill — speed up client delivery by turning every call into a clean action list.
• SEO Long-form Writer skill — for when your side business is ready for inbound content instead of pure outreach.
• Deep Research Brief skill — research any niche or buyer before you pitch, so your outreach lands informed.

If this was useful, I send one practical, hype-free workflow like this every week. Join the newsletter at alexandrrich.substack.com.