12 Ways to Make Money with Claude, Ranked
Twelve realistic ways professionals earn with Claude — ranked from fastest first dollar to most scalable — each with the exact workflow and honest difficulty.
Alexandr Rich
AI Made Practical
I've spent the last couple of years watching people try to make money with AI, and I've noticed something uncomfortable: most of the advice online is written by people who make money talking about making money with AI. The actual paths — the ones where a normal person with a job and no coding background earns real dollars — are quieter, slower, and far more repeatable than the hype suggests.
So this post is my attempt to fix that. Below are twelve ways to earn money with Claude that I consider genuinely realistic. Not "become a millionaire" realistic. "Earn your first $200 this month, and maybe build it into $2,000 a month over a year" realistic.
Two things make this list different from the usual listicles:
- Every idea comes with a workflow, not a slogan. For each path you'll get who it fits, an exact first-week plan, the actual Claude prompts to use, pricing guidance, and an honest difficulty rating. You can start any of these tonight.
- The ranking is honest about the trade-off nobody mentions. The fastest first dollar and the biggest long-term income are almost never the same path. So I've ranked these from first dollar to most scalable — the list starts with things that can pay you next week and ends with things that could pay you while you sleep, eventually, after real work.
One rule before we start: Claude is a lever, not a magic wand. Every path below still requires you to find a customer, deliver something they value, and do it again. Claude makes each of those steps 3–10x faster. It does not make them optional.
Let's go.
1. Freelance writing and editing, upgraded with Claude
Speed to first dollar: days. Scalability: low. Difficulty: 2/5.
This is the most boring idea on the list, which is exactly why it works. There is a functioning market for writing and editing — blog posts, web copy, case studies, email sequences, white papers — with existing buyers, existing platforms, and existing rates. You are not inventing demand. You are entering a market with a productivity advantage.
Who it fits: anyone who writes reasonably well in their native language and can tell good writing from bad. You don't need to be a great writer. You need to be a great editor of Claude's writing, which is a much lower bar.
The honest catch: rates for generic "content writing" have fallen because everyone has AI now. The money has moved to people who bring judgment: industry knowledge, interviewing skills, and the ability to make a piece sound like a specific company rather than like AI. That's the job now. Claude does the typing; you do the thinking.
First-week plan:
- Day 1–2: Pick a niche you actually know from your day job (logistics, HR, dentistry, accounting — anything). Write two sample articles in that niche using the workflow below. Publish them somewhere public: LinkedIn, a free blog, anywhere with a URL.
- Day 3–4: Set up profiles on two freelance marketplaces and send 10 personalized pitches to small companies in your niche whose blogs have gone quiet (a dead blog is a buying signal — someone decided content matters, then ran out of time).
- Day 5–7: Send 10 more pitches. Respond fast to anything that comes back. Offer a paid trial article at a modest rate to lower the risk of saying yes to you.
The Claude workflow: the core skill is separating research, drafting, and voice into different steps instead of asking for "an article" in one prompt.
Prompt 1 — the brief:
You are helping me write an article for [company], which sells [product] to [audience]. Before writing anything, ask me 8 questions whose answers would make this article impossible for a generic AI writer to produce — questions about the company's actual customers, real examples, internal data, and opinions. Number them so I can answer inline.
Answer those questions from your own knowledge or a 15-minute call with the client. This step is where the entire value lives.
Prompt 2 — the draft:
Using my answers above, write a 1,500-word article titled [title]. Structure: a specific opening scenario (no throat-clearing), 3–4 substantive sections with concrete examples from my answers, one honest section about limitations or trade-offs, and a short conclusion with one clear next step. Plain English, short sentences, no hype words like "unlock," "supercharge," or "game-changer."
Prompt 3 — the voice pass:
Here are three paragraphs the client wrote themselves: [paste from their site or emails]. Rewrite the draft to match this voice — sentence length, formality, vocabulary. Then list the 5 biggest changes you made so I can check them.
Pricing: don't charge by the word; charge by the piece. $150–$400 per article is a realistic starting range for niche B2B content, rising to $500+ once you have results to point at. Your Claude-assisted cost is 2–3 hours per piece including client communication.
2. Resume and cover letter services
Speed to first dollar: days. Scalability: low. Difficulty: 1/5.
The fastest first dollar on this list, full stop. Job seekers are motivated, the deliverable is small, and the turnaround is same-day. This is also the lowest-ceiling business here — you will never scale a resume service into a big income without turning it into something else. But as a confidence-builder and a first taste of getting paid for AI-assisted work, nothing beats it.
Who it fits: anyone who has hired people, worked in HR or recruiting, or simply reviewed a lot of resumes. If you've sat on the other side of the table in any capacity, you have the judgment this service needs.
First-week plan:
- Day 1: Rewrite your own resume and one friend's resume using the workflow below. These are your before/after samples.
- Day 2–3: Post in your own network (LinkedIn, community groups): "I'm doing 5 resume rewrites this week at an introductory price of $X. Reply and I'll send details." Warm networks convert far better than marketplaces for this service.
- Day 4–7: Deliver fast, ask each happy customer for a testimonial and one referral. Referrals are the whole engine of this business.
The Claude workflow:
Prompt 1 — extraction:
Here is my client's current resume and the job posting they're targeting: [paste both]. First, list every requirement in the posting. Second, for each requirement, find evidence in the resume that addresses it, or mark it MISSING. Third, give me 6 interview-style questions I should ask my client to surface accomplishments the resume leaves out — especially numbers: team sizes, budgets, percentages, timelines.
Send those questions to the client. Their answers are the raw material generic AI tools never get.
Prompt 2 — the rewrite:
Using the resume, the job posting, and my client's answers, rewrite the resume. Rules: every bullet starts with a strong verb and includes a concrete number or outcome where one exists; mirror the posting's key terminology naturally (no keyword stuffing); cut anything that doesn't support this specific application; keep it to [1/2] pages. Then write a cover letter under 300 words that tells one specific story from their answers and connects it to the company's stated need.
Pricing: $75–$150 for resume plus cover letter as a beginner; $200–$400 once you specialize in an industry or seniority level. Each job takes you 60–90 minutes. Specialization is what raises the price — "resumes for nurses moving into administration" beats "resumes" every time.
3. Meeting notes and documentation services for small firms
Speed to first dollar: 1–2 weeks. Scalability: medium. Difficulty: 2/5.
Small professional firms — law offices, accounting practices, contractors, clinics, agencies — drown in undocumented knowledge. Meetings happen, decisions evaporate. Processes live in one employee's head. They know it's a problem; they will never fix it themselves. You can sell the fix as a recurring service, and recurring is the magic word: this is the first idea on the list that produces predictable monthly income.
Who it fits: organized people. Ex-admins, project managers, operations folks, executive assistants. If you're the person who sends the follow-up email nobody asked for but everyone uses, this is yours.
First-week plan:
- Day 1–2: Pick one service: meeting minutes (they record, you deliver structured notes and action items within 24 hours) or process documentation (you interview a staff member, deliver a step-by-step SOP). Create one sample of each from a real or realistic scenario.
- Day 3–5: List 20 local small firms. Email or call the office manager — not the owner — with a one-line offer: "I turn your recorded meetings into decisions, action items, and deadlines, delivered next morning. First meeting free."
- Day 6–7: Deliver the free samples fast and beautifully. Then quote the monthly retainer.
The Claude workflow: for meeting notes, get a transcript (most video tools produce one; a phone recording plus a transcription app also works), then:
Here is a meeting transcript: [paste]. Produce: (1) a 5-sentence summary a director could read in 60 seconds; (2) every decision made, with who made it; (3) every action item as a table — owner, task, deadline (write "no deadline set" where none was given, don't invent one); (4) open questions that were raised but not resolved; (5) anything that sounded like a commitment to a client or vendor, flagged separately. If a name is unclear in the transcript, mark it [UNCLEAR] rather than guessing.
For SOPs, record a 20-minute interview with the person who does the task, then:
Turn this interview transcript into a step-by-step standard operating procedure titled "[task name]." Number every step. For each step include: what to do, what tool or screen it happens in, and what "done" looks like. Add a "Common mistakes" section from anything the interviewee warned about. Flag any step where the transcript was ambiguous with [VERIFY] so I can check it with the client.
That [VERIFY] habit matters. Your clients are trusting you with their operations; the fastest way to lose them is a confidently wrong document.
Pricing: meeting notes at $200–$500/month per firm for a weekly meeting cadence; SOPs at $150–$300 per documented process. A single person can comfortably serve 8–12 small firms — that's a real income from work most people find tedious and you, with Claude, will find fast.
4. Research briefs for agencies and consultants
Speed to first dollar: 1–3 weeks. Scalability: medium. Difficulty: 3/5.
Marketing agencies, PR firms, and consultants constantly need background research: market overviews, competitor teardowns, prospect briefings before a pitch, industry summaries for a client deck. This work is billable to their clients, urgent, and annoying for them to staff. Selling research to businesses that resell it is one of the most underrated freelance niches I know, because your buyer is a repeat buyer by definition.
Who it fits: analytical people who enjoy synthesizing information — ex-journalists, analysts, grad students, librarians at heart. Strong reading comprehension and healthy skepticism are the core skills.
The honest catch: the difficulty rating is 3/5 for one reason — verification. Claude can hallucinate statistics and sources, and one fabricated number delivered to an agency destroys the relationship. Your process must assume every specific claim is wrong until you've checked it.
First-week plan:
- Day 1–2: Produce one killer sample: a 5-page competitor teardown of a well-known brand in a niche you understand. Make it look professional — clean structure, cited sources, an executive summary.
- Day 3–5: Identify 15 small agencies (5–30 people). Email the founder: "I produce pitch-ready research briefs — competitor teardowns, market overviews, prospect briefings — with 48-hour turnaround. Here's a sample. Flat rate per brief."
- Day 6–7: Follow up. Agencies are busy; a polite second email doubles response rates.
The Claude workflow: three passes — structure, synthesis, verification.
Prompt 1 — structure:
I'm producing a research brief on [topic] for a marketing agency pitching [type of client]. Draft the brief's skeleton: section headings, the specific questions each section must answer, and for each question, what kind of source would credibly answer it. Prioritize what a pitch team actually needs over academic completeness.
Then gather real sources yourself — company sites, filings, industry reports, news. Paste the relevant excerpts into the conversation.
Prompt 2 — synthesis:
Using ONLY the source material I've pasted above, fill in the brief. Every factual claim must reference which source it came from, inline, like [S3]. Where the sources conflict, say so. Where a section question can't be answered from these sources, write "Not established by available sources" instead of filling the gap. End with a "What this means for the pitch" section — 5 implications, clearly labeled as interpretation.
Prompt 3 — the adversarial check:
Now act as a skeptical fact-checker reviewing this brief. List the 10 claims most likely to be wrong, outdated, or overstated, and for each, tell me exactly what I should verify before sending this to a client.
Check those ten things by hand. This last pass is 30 minutes that protects your entire reputation.
Pricing: $250–$600 per brief depending on depth; $1,000+ for full market overviews. Push toward a retainer after the third order: "4 briefs a month for $1,400" is a sentence that changes your financial life.
5. AI-assisted consulting in your existing profession
Speed to first dollar: 2–4 weeks. Scalability: medium. Difficulty: 2/5.
Here's the idea most people overlook because it's hiding in plain sight: don't start a new career — add an AI layer to the one you already have. If you're an accountant, marketer, teacher, recruiter, project manager, or paralegal, you possess the two things AI can't fake: domain judgment and trust from people who already know you. Claude turns your existing expertise into deliverables at three times the speed, which means you can consult on the side without it eating your life.
Who it fits: anyone with 3+ years in a profession and at least a small network in it. This is the best path on this list for most employed professionals, and the one I recommend when someone asks "where do I start?"
First-week plan:
- Day 1–2: Write down the five questions people at work already come to you for. Those are your services. Pick the one with the clearest deliverable (an audit, a plan, a review, a template set).
- Day 3–4: Build the deliverable once, for a fictional client, using Claude. Time yourself. This tells you your real cost and gives you a sample.
- Day 5–7: Tell 10 people in your network — former colleagues, industry contacts — exactly what you now offer and the fixed price. Not "I'm doing consulting now" (too vague to buy) but "I do [specific deliverable] for [specific type of company] at [price]."
The Claude workflow: the pattern is your judgment in, structured deliverable out. Example for a marketing professional offering website audits:
I'm a marketing consultant auditing the website of [company], which sells [product] to [audience]. Here is the text of their homepage, pricing page, and about page: [paste]. I'll give you my rough observations as messy notes: [your notes]. Turn my observations into a professional audit document: an executive summary, findings grouped by theme, each finding with (a) what I observed, (b) why it costs them money, (c) a specific fix, ranked by effort vs. impact. Keep my opinions as the spine — organize and sharpen them, don't replace them.
Notice the structure: Claude is formatting and sharpening your expertise, not substituting for it. That's why clients pay consultant rates rather than AI rates.
A second prompt that earns its keep in any profession:
Here is my draft deliverable. Act as the skeptical client who's about to pay for this. What would they push back on? What's vague? What question will they ask in the review meeting that I'm not ready for? Give me the 5 hardest ones with suggested answers.
Pricing: whatever your profession bills, at the freelance rate for your market — typically $75–$200/hour, but packaged as fixed-price deliverables ($500 audit, $1,500 plan) so your Claude-driven speed becomes margin instead of a discount. Never sell hours when AI is doing the acceleration; you'd be handing the productivity gain to the client.
6. Productized reports
Speed to first dollar: 3–6 weeks. Scalability: medium-high. Difficulty: 3/5.
A productized report is a fixed-scope, fixed-price document you produce the same way every time: a local market report for realtors, a monthly competitor digest for e-commerce brands, an SEO content-gap report, a quarterly industry snapshot for a niche trade. It's idea #4 (research briefs) evolved from custom work into a product — same input slot, same process, same output format, different customer each time. That standardization is what makes it scale: your twentieth report takes a quarter of the time your first one did.
Who it fits: people who've done a few custom research or consulting jobs (paths 4 and 5) and noticed themselves producing the same thing repeatedly. Productizing too early — before you've validated that anyone pays for the custom version — is the classic mistake here.
First-week plan:
- Day 1–3: Take the deliverable you've made most often (or the one your industry obviously lacks) and define it rigidly: exact sections, exact length, exact inputs required from the buyer, exact price, exact turnaround. Write this on a simple one-page description.
- Day 4–5: Build the master prompt (below) and produce two full samples end-to-end. Fix everything that annoyed you.
- Day 6–7: Offer it to 15 prospects at a founding-customer price. The pitch is short because the product is defined: "I produce [report] for [niche]. Fixed price [X], delivered in [Y] days. Here's a sample."
The Claude workflow: the asset you're really building is a reusable master prompt — a template where only the client-specific inputs change:
You are producing the [Report Name], a standardized report I sell to [niche]. Follow this exact structure: [paste your locked section list, with a 2-line description of what each section must contain and its target length]. Here are this client's inputs: [paste the same categories of input you collect every time — their data, their market, their competitors' pages]. Rules: use only the inputs provided; mark anything unverifiable as [CHECK]; match the tone of this sample report: [paste one strong past example]. Produce the full report.
Keep that master prompt in a document and improve it after every delivery. Version 10 will be dramatically better than version 1, and it's an asset a competitor can't see. When you're ready to go further, turn it into a proper skill file — I've written about that in my productized AI service skill.
Pricing: $99–$500 per report depending on the niche's budget, with a subscription version ($X/month for a monthly refresh) as the real prize. Ten subscribers at $250/month is $2,500 of largely systematized monthly income.
7. Content repurposing service
Speed to first dollar: 1–3 weeks. Scalability: medium-high. Difficulty: 2/5.
Every founder, coach, and expert with a podcast, YouTube channel, or webinar library is sitting on a pile of content they know they should be turning into LinkedIn posts, newsletters, blog articles, and clips scripts — and they aren't, because it's a grind. You sell the grind: "Send me your weekly episode, get back 10 pieces of derived content in 48 hours." It's recurring, the raw material is handed to you, and the client's voice is right there in the transcript for Claude to learn from.
Who it fits: people with a feel for social content — you know a scroll-stopping first line from a dead one. Light marketing instinct required; heavy writing skill not.
First-week plan:
- Day 1–2: Pick a niche (business podcasts, fitness coaches, real-estate YouTubers). Take one public episode from a creator you'd love as a client and produce a full repurposing pack: 5 LinkedIn posts, 1 newsletter edition, 1 blog article, 3 short-video scripts.
- Day 3–5: Send that pack to the creator, free, no strings: "I made this from episode 47. If you'd like this every week, here's my rate." Then do the same for four more creators. This "free spec work to the exact buyer" pitch converts unusually well because the product is sitting in their inbox.
- Day 6–7: Follow up, refine the pack based on any feedback.
The Claude workflow:
Prompt 1 — voice capture (once per client):
Here are three pieces this creator wrote or said in their own words: [paste transcript excerpts and any posts they've written]. Describe their voice as a style guide I can reuse: sentence length, energy, vocabulary they love, vocabulary they'd never use, how they open, how they close, their stance (teacher? provocateur? friendly peer?).
Prompt 2 — extraction:
Here is this week's episode transcript: [paste]. List the 12 strongest atomic ideas in it — a specific claim, story, number, or contrarian take per idea. For each, quote the exact lines from the transcript that support it. Rank them by how well they'd stop a scroll.
Prompt 3 — production:
Take ideas 1–5 and write one LinkedIn post for each, following the style guide from earlier. Rules: first line must create curiosity or tension without clickbait; one idea per post; use the creator's actual words from the transcript quotes where they're strong; end with a question or a soft pointer to the full episode. Then turn ideas 1–3 into a 600-word newsletter edition with a single narrative thread.
Pricing: $500–$1,500/month per client depending on volume. Each weekly pack takes 2–4 hours once your prompts are tuned. Five clients is a serious side income; ten is a job replacement, and at that point you can hire your first helper because the process is already written down.
8. Building and selling prompt packs and skill files
Speed to first dollar: 4–8 weeks. Scalability: high. Difficulty: 3/5.
Everything above sells your time, accelerated. From here on, the list shifts to selling things — and the first thing you can sell is your workflows themselves. A prompt pack is a set of battle-tested prompts for a specific job ("30 prompts for property managers"). A skill file is the more powerful evolution: a structured document that teaches Claude an entire repeatable process — inputs, steps, quality checks, output format — so the buyer gets your whole workflow, not just openers. I publish several on this site, like my cover letter writer and deep research brief skills, so you can see the format in the wild.
Who it fits: people who have already used Claude for real work — ideally one of paths 1–7 — and have prompts that are demonstrably better than what a beginner writes. Selling prompts you haven't used to earn money yourself is how you end up with a product nobody needs.
The honest catch: the product takes days, but the audience takes months. Digital products don't sell themselves; they sell to people who already trust you. That's why this is ranked 8th despite being cheap to build.
First-week plan (product week — the audience takes longer):
- Day 1–2: Choose one narrow job for one narrow buyer. "Prompts for writing" is dead on arrival; "the complete Claude workflow for insurance agents writing renewal emails" can charge real money.
- Day 3–5: Write 15–25 prompts or one full skill file. For each prompt: what it's for, the prompt itself, what to customize (marked in [brackets]), and a real example of the output.
- Day 6–7: Test every single one in a fresh conversation. Cut the weak third. A small pack that all works beats a big pack that half works.
The Claude workflow: yes, Claude helps you build the product about Claude:
I'm creating a prompt pack for [specific buyer] doing [specific job]. Here are 6 prompts I've actually used successfully, with notes on when each worked: [paste]. First, critique each one — what makes it work, where it's fragile, what a novice would break when customizing it. Then help me generalize each into a template with [BRACKETED] slots and a one-line usage note. Finally, list 10 more tasks this buyer does weekly that deserve a prompt, ranked by pain.
And to pressure-test quality:
Act as a skeptical customer who paid $39 for this pack. Go through it and tell me: which prompts feel like filler, which instructions are ambiguous, and what's missing that you'd have expected for the money.
Pricing: $19–$49 for a focused prompt pack, $49–$149 for a full skill file or workflow system, sold through a simple storefront (Gumroad-style platforms make this a one-evening setup). Expect early sales to come from your own content and network — which is exactly why path 10 (the newsletter) pairs with this one.
9. Teaching AI workshops at work and beyond
Speed to first dollar: 4–8 weeks. Scalability: medium-high. Difficulty: 3/5.
Every company is somewhere on the awkward curve between "we should be using AI" and "nobody actually knows how." If you become the person who can stand in front of a team and show them — concretely, with their own tasks — how to use Claude well, you have a service that commands hundreds to thousands of dollars per session. And the best first venue is the company you already work for.
Who it fits: people comfortable presenting, or willing to get comfortable. You do not need to be an AI expert. You need to be two chapters ahead of the audience and fluent in their work. A bookkeeper teaching bookkeepers beats an AI researcher teaching bookkeepers every time.
First-week plan:
- Day 1–3: Build one 90-minute workshop for one job function you know well: "Claude for project managers," "AI for the sales team." Structure: 20 minutes of principles, 60 minutes of hands-on exercises using their real task types, 10 minutes of "what to do this week."
- Day 4–5: Pitch it internally — your manager, your L&D team: "I've built a hands-on AI workshop for our team. Can I run it at the next team meeting?" Internal first: it's a paid rehearsal (you're on salary), it's low-stakes, and it produces a reference.
- Day 6–7: Run it, collect written feedback, and ask the question that opens the next door: "Do you know another team — or another company — that would want this?"
The Claude workflow: use Claude to build the curriculum and the exercises:
I'm designing a 90-minute hands-on Claude workshop for [role] at [type of company]. Their typical weekly tasks: [list]. Design the session: (1) three core principles worth teaching, each in one sentence; (2) five hands-on exercises using realistic versions of their tasks, ordered easy to hard, each with a starter prompt, the expected "aha," and the mistake most beginners will make; (3) a one-page handout of prompts they can use the next morning; (4) the three questions skeptics in the room will ask, with honest answers.
The honest answers matter. The credibility moment in every AI workshop I've seen is when the presenter says "here's where it fails" — show a hallucination live, show a bad prompt and a good one side by side. Hype loses rooms; honesty wins repeat bookings.
Pricing: internal workshops build the asset; external ones monetize it. $500–$1,500 per session for small businesses, $2,000–$5,000 for corporate half-days once you have testimonials. Scalability comes from repetition — the same workshop delivered twenty times — and eventually from recording it as a course.
10. A niche newsletter
Speed to first dollar: 3–6 months. Scalability: high. Difficulty: 4/5.
Now we're firmly in "slow first dollar, compounding upside" territory. A newsletter is not a quick income path — anyone who tells you otherwise is selling a course about newsletters. What it is: the single best long-term asset on this list, because an audience that trusts you makes every other path easier. Your prompt packs (path 8), workshops (path 9), and services (paths 1–7) all sell better to a list that already reads you weekly.
Who it fits: people with genuine, durable interest in a specific niche — because you'll be writing about it in week 40 whether or not it's fun that day. Consistency beats brilliance in this game, by a wide margin.
The honest catch: the median newsletter dies at issue 6. The failure mode isn't quality; it's quitting. Decide up front that you're writing 26 issues before you judge the results.
First-week plan:
- Day 1–2: Pick a niche at the intersection of what you know, what you'd read anyway, and what some group spends money on ("AI for independent insurance agents" — narrow, monetizable — over "AI news" — broad, worthless).
- Day 3–4: Define a repeatable format: e.g., one practical workflow, one tool or idea tested honestly, three curated links with your one-line take. Format is what makes issue 30 writable on a tired Tuesday.
- Day 5–7: Write issues 1 and 2 with the workflow below, set up on a newsletter platform, and get your first 20 subscribers by personally asking people who'd genuinely benefit. Yes, personally. Everyone starts at zero.
The Claude workflow: Claude is your editor and researcher here — not your ghostwriter. Newsletters live on voice, and readers unsubscribe from generic AI prose within two issues.
Research assistant:
Here are my raw notes and links for this week's issue on [topic]: [paste]. Help me find the angle: what's the one non-obvious point in these notes? What would my reader — a [description] — actually change in their week after reading this? Give me 3 candidate angles with a first line for each.
Editor, after you write the draft:
Here's my draft. Edit it like a sharp newsletter editor: cut 20% of the words without losing meaning, flag every sentence that sounds like generic AI writing, tighten the opening to earn the next paragraph, and check that the practical takeaway is concrete enough to act on tomorrow. Show edits as a list so I keep my own voice.
Pricing and monetization: sponsorships become viable around 1,000–2,000 engaged niche subscribers ($100–$500 per issue in a commercial niche); your own products and services monetize at any size; paid subscriptions ($5–$10/month) work once you consistently deliver something readers can't get elsewhere. Realistic timeline to meaningful money: 6–12 months of weekly consistency.
11. AI-assisted micro-SaaS with Claude Code
Speed to first dollar: 2–6 months. Scalability: very high. Difficulty: 5/5.
Here's the path that genuinely wasn't available to non-programmers three years ago: building a small, paid software product with Claude Code — Anthropic's tool that writes, runs, and fixes code from plain-English instructions. I want to be very careful here, because this is where hype does the most damage. The truth: a non-technical person can now ship a simple working web tool. A non-technical person cannot skip the hard parts — knowing what to build, getting anyone to use it, and maintaining it when it breaks. Software was never hard because of typing code; it's hard because of everything around the code. Claude removes the typing.
Who it fits: patient, systematic people who already know a niche's painful problem firsthand — ideally from paths 3–7, where clients keep asking for the same thing. "I keep doing this manually for clients; what if it were a tool?" is the single best origin story for micro-SaaS.
The honest catch: rated 5/5 difficulty not because any single day is hard, but because the failure modes are invisible until month three. Scope creep, building without customers, and abandoning at the first confusing bug kill most attempts. Build the smallest thing that someone will pay for, and talk to that someone before you write a line.
First-week plan (validation, not building):
- Day 1–3: Write down the manual process you or your clients repeat. Describe the tool in one sentence: "It takes [input] and produces [output] for [who]." If you can't, you're not ready to build.
- Day 4–5: Pitch the one-sentence tool to 10 people who have the problem. Ask one question: "Would you pay $X/month for this — and can I show you a rough version in three weeks?" Three yeses means build. Zero means thank Claude for the weeks it just saved you.
- Day 6–7: With yeses in hand, install Claude Code and build the "walking skeleton": the ugliest possible version that does the core transformation once, end to end.
The Claude workflow: the skill is describing behavior precisely and working in small steps.
The spec prompt (in Claude, before Claude Code):
I'm a non-programmer planning a small web tool: [one-sentence description]. Interview me one question at a time to produce a plain-English spec: what the user sees, what they input, what happens, what can go wrong, what "working" means. Then write the spec as a numbered list of small, testable behaviors, ordered so each builds on the last. Flag which items are essential for version 1 and which I should ruthlessly postpone.
Then, in Claude Code, feed it one behavior at a time:
Build behavior #1 from this spec: [paste it]. Keep it as simple as possible — no features I didn't ask for. When done, tell me exactly how to run it and what I should see if it works.
And the prompt that saves non-programmers' sanity:
It broke. Here's what I did, what I expected, and what happened instead: [describe, paste any error text]. Explain what went wrong in plain English before fixing it, so I learn what this failure looks like next time.
Pricing: $10–$50/month per user for a niche tool. The realistic early outcome isn't a startup — it's 20–100 subscribers, $500–$3,000/month, from a tool serving a niche you know. That's a life-changing side income, and some of the best businesses I've seen started exactly this small.
12. Done-for-you automation setup
Speed to first dollar: 1–3 months. Scalability: very high. Difficulty: 4/5.
The most scalable service business on this list, and where several earlier paths converge. Small businesses don't want to learn AI — they want the outcome: leads automatically triaged, inquiries drafted for reply, meeting notes filed, weekly reports generated. You come in, map one painful workflow, wire it up (Claude plus off-the-shelf automation platforms that connect apps without code), charge for the setup, and then — the crucial part — charge monthly to keep it running. Setup fees pay you once; the maintenance retainer is what stacks into a real business, because automations serve clients while you sleep, and each new client adds recurring revenue without adding proportional hours.
Who it fits: process-minded tinkerers. If you've enjoyed paths 3, 5, or 6 and find yourself thinking "why does a human do this step at all?", you're the right person. You'll need a few weekends learning a no-code automation platform — genuinely learnable without a technical background, but not free.
First-week plan:
- Day 1–3: Automate something of your own, end to end: e.g., "new email inquiry → Claude drafts a reply in my voice → draft lands in my drafts folder for review." Feeling a whole pipeline work is the education.
- Day 4–5: Write up that automation as a one-page case study: the before (minutes per day, errors, delays), the after, and a screenshot.
- Day 6–7: Offer a free "automation audit" to five small businesses you know: a 30-minute call where you map their most repetitive workflow and quote a fixed price to automate it. The audit is your sales process and your requirements-gathering in one.
The Claude workflow: two layers — Claude helps you design the automation, and Claude sits inside the automation doing the language work.
Design layer:
I'm automating this workflow for a client: [describe current manual process, step by step]. Break it into steps and classify each: (a) pure data movement a no-code platform handles, (b) judgment/language work for an AI step, (c) things that must stay human. For each AI step, draft the instruction prompt, including explicit rules for when it should refuse to act and route to a human instead. Then list the top 5 ways this automation could fail embarrassingly, and a safeguard for each.
Inside the automation, every Claude step needs a tightly scoped prompt — this is what separates professional setups from dangerous ones:
You are a step in an automated workflow for [company]. Your only job: read the incoming [email/form/document] below and produce [exact output format]. Rules: if the message mentions [refunds/legal issues/anger/anything ambiguous], output ESCALATE and nothing else. Never invent information not present in the input. Output must follow this exact structure: [template].
Always build in a human-review stage at first (drafts, not sends). Graduate to full automation only after weeks of clean output — and tell clients that's your standard, because it should be.
Pricing: $1,000–$5,000 per automation setup depending on complexity, plus $200–$500/month per client for monitoring, fixes, and improvements. Fifteen maintenance clients at $300/month is $4,500 of recurring revenue on top of new setup fees. This is the closest thing on this list to building an agency, and it's how several one-person AI businesses I respect actually make their living.
The ranking at a glance
| # | Path | First dollar | Scalability | Difficulty | Best for |
|---|---|---|---|---|---|
| 1 | Freelance writing/editing with Claude | Days | Low | 2/5 | Decent writers entering an existing market |
| 2 | Resume & cover letter services | Days | Low | 1/5 | Anyone who's hired or reviewed candidates |
| 3 | Meeting notes & documentation | 1–2 weeks | Medium | 2/5 | Organized ex-admins, ops people |
| 4 | Research briefs for agencies | 1–3 weeks | Medium | 3/5 | Analytical readers, careful checkers |
| 5 | AI-assisted consulting in your field | 2–4 weeks | Medium | 2/5 | Professionals with 3+ years' experience |
| 6 | Productized reports | 3–6 weeks | Medium-high | 3/5 | People repeating the same custom work |
| 7 | Content repurposing service | 1–3 weeks | Medium-high | 2/5 | Social-content-savvy grinders |
| 8 | Prompt packs & skill files | 4–8 weeks | High | 3/5 | Practitioners with proven workflows |
| 9 | Teaching workshops | 4–8 weeks | Medium-high | 3/5 | Confident presenters, patient explainers |
| 10 | Niche newsletter | 3–6 months | High | 4/5 | Consistent writers with a durable niche |
| 11 | Micro-SaaS with Claude Code | 2–6 months | Very high | 5/5 | Patient builders who know a niche's pain |
| 12 | Done-for-you automation setup | 1–3 months | Very high | 4/5 | Process-minded tinkerers |
How to actually choose
If you've read this far, the risk isn't picking the wrong path — it's picking three and starting none. So here's my honest advice:
If you need money soon: start with #1, #2, or #3. They're unglamorous, low-ceiling, and they work. A first paid invoice changes your psychology more than any course ever will.
If you have a profession: #5 is almost certainly your best hourly return. You already have the trust and the judgment; Claude just gives you the speed.
If you're playing a longer game: run a "now and later" pair. One fast path for income and learning (say, #4 or #7), plus one compounding asset you feed weekly (#10, or the systematization that leads to #6, #8, or #12). Every durable AI income story I know follows this shape: service work funds the ramp, the asset becomes the business.
Whatever you pick: do it for six weeks before judging it. Every path above looks like it isn't working at week two. That's not a signal; that's just week two.
And keep the standard high. The people making real money with Claude in 2026 aren't the ones with secret prompts — the prompts are right here in this post. They're the ones who check facts before sending, hit deadlines, say "I don't know" when they don't, and treat AI as a lever on their judgment rather than a replacement for it. Boring, I know. Boring is what clients pay for twice.
Resources & links
- /blog/make-money-with-ai-2026 — my big-picture guide to earning with AI this year, if you want the strategy layer above this tactics post.
- /blog/three-chatgpt-prompts-make-money-fast — three prompts that generate income ideas fast, a good warm-up before committing to a path.
- /blog/one-person-business-with-ai-30-days — a 30-day plan for turning one of these paths into a real one-person business.
- /blog/claude-workflow-template — the reusable workflow template behind every prompt sequence in this post.
- /blog/write-seo-articles-with-ai — my full process for AI-assisted articles, useful for paths 1, 7, and 10.
- /skills/productized-ai-service — a downloadable skill for packaging a repeatable service like the productized reports in path 6.
- /skills/cover-letter-writer — the exact skill file behind path 2's cover letter workflow.
- /skills/seo-longform-writer — my long-form writing skill, the heavy-duty version of path 1's article workflow.
- /skills/meeting-notes-to-actions — the meeting-transcript-to-action-items skill that powers path 3.
- /skills/deep-research-brief — the research skill behind path 4, with the verification pass built in.
- /skills/ai-marketing-team — a skill that turns Claude into your one-person marketing department, handy for promoting whichever path you choose.
If this was useful, I send one practical, hype-free AI workflow like this every week — join me at https://alexandrrich.substack.com.
Download the skills from this guide
Put the ideas above into practice — grab these ready-to-use AI workflows.
