How to Use Claude to Build a $3K/Month Side Project (Real)

Most working adults assume Claude only pays off if you already run an AI agency, sell a Claude course, or live in San Francisco with a venture firm on speed dial. That assumption is wrong, and the proof is sitting on a public Indie Hackers page right now, with the receipts open for anyone with a browser tab to verify.

This post is the long-form companion to my video on a single Chrome extension developer named Edmund Yong, who quietly used Claude to build one product clearing $3,700 a month in monthly recurring revenue and $42,000 in total revenue at the six-month mark. His Indie Hackers page is public. His revenue chart is public. His GitHub repo full of Claude commands is public. You can open all three in a different tab while you read this.

You’ll walk out of this post with the math, the three-step pattern Edmund used, the part most “make money with Claude” videos skip, and the one tactical move you can lift on a Saturday morning even if your day-job skill has nothing to do with code. I built this to serve search traffic that hasn’t seen the YouTube version yet, so if you’d rather watch than read, here it is.

[VIDEO_EMBED: https://notes.platformproof.com/notes/he-quietly-used-claude-3k-month]

If you want the one-page Action Notes worksheet to follow along, it’s right here.

What you’ll learn

The verified math behind a $3K/month Claude side project (no agency, no audience, no course)
Why Claude shrinks your TIME, not your skill, and what that flips for working adults
The four-question test that tells you whether your desk-job skill is sellable
The three-step pattern Edmund used to go from $0 to $3,700/mo over six months
The one tactical move sitting on his public GitHub repo that nobody else doing this is talking about
Where to ship a digital product in 2026 (Gumroad and your own site beat the alternatives)
How to stop chasing AI-agency content and start building something on the skill you already have

The real receipt: $3,700/mo MRR on a single Chrome extension

Edmund Yong runs a Chrome extension called Easy Folders. It adds folders, search, and a prompts manager to Claude.ai and ChatGPT. It is one product. It has one founder. He has a single public product page on Indie Hackers that anyone on the internet can open in another tab.

At the six-month mark after launch he posted a public milestone on that same page reporting $3,700 in monthly recurring revenue and $42,000 in total revenue. That milestone post is still live. His revenue chart sits on the product’s revenue page. His founder profile is at indiehackers.com/edmundd. His X account is here. His GitHub profile is here.

I want to be specific about why this matters. Most “make money with Claude” content online is invented. In the first week of researching this video I disqualified fourteen case studies. Some had numbers that didn’t match the dates. Some had product pages that 404’d when I went to check. Some were obvious AI-generated stories dressed up as people. Edmund’s math was the only one that survived. The page is public. The chart is public. The GitHub repo is public. Open it. Verify it. Then come back.

The math behind the $3K/month number

The math is small enough to feel boring. That’s the point.

Roughly $25 per sale
Roughly 120 paid users a month
$25 × 120 = $3,000 a month from the core product price band

That’s four sales a day. It is not viral. It is not luck. It is not a course about courses. It is one product, one skill, one listing, and one founder running a small Chrome extension business that scales because the product solves a real problem for Claude.ai and ChatGPT users who organize a lot of chats.

The reason that math matters for a working adult is simple. Four sales a day is something you can imagine. It is not a Shopify dropshipping fantasy or a SaaS unicorn pitch deck. It is a number a person with a day job and two kids can model on the back of an envelope and believe.

Why Claude shrinks the time, not the skill

Most of the AI content swimming around YouTube right now gets the relationship between you and Claude exactly backward. They sell Claude like it’s the skill. Like learning Claude is the path. Like prompt engineering is the new welding.

That framing is wrong on its face, and Edmund is the proof.

Edmund had the Chrome extension build skill before Claude existed. He’d been working on browser extensions and tooling at his day job for years. The pattern of “ship a small Chrome extension that solves one annoying problem” lived in his head and on his hard drive long before Claude was downloadable. Claude did not give him the skill. Claude shortened the runway between the idea and the shipped product.

If you took Claude away from Edmund tomorrow, he would still have the skill. He would just be a slower version of himself. The hours of drafting code, writing the FAQ, building the description, formatting the changelog, drafting the support docs — Claude turns that into a Saturday morning instead of a Saturday month. Same product. Different time.

That distinction matters because it changes who the opportunity is for. If Claude is the skill, the opportunity belongs to AI developers and Claude course sellers. If Claude is the leverage on a skill you already have, the opportunity belongs to anyone with a desk job and a Saturday.

The skill you’ve stopped seeing

You know the kind of skill you’ve stopped seeing as a skill. The spreadsheet you’ve built and rebuilt at work over four years. The SOP you’ve maintained so long nobody else on the team knows how it actually works. The HR onboarding checklist that exists only because you wrote it three jobs ago and brought it with you. The vendor-management process. The quoting template. The post-incident report format.

That’s where Claude’s leverage hides. Not in learning Claude. In applying Claude to the thing you already do. And the reason this matters is simple — the people on YouTube selling Claude courses make money from your confusion about the tool. They do not make money from you noticing the skill that’s already sitting on your laptop.

The 4-question test for spotting a sellable skill

Before you spend a weekend in Claude building anything, you need to know which skill to point Claude at. Most people skip this step and waste two months building a product on the wrong thing. Here is the test I run with people:

1. Has a coworker walked over to your desk in the last six months and asked you to explain something more than once? Not once. More than once. That’s the moment your way became different from the default way. That’s the moment a skill became sellable. 2. When something in this area breaks at work, do people text you first instead of the manager? If yes, you’re the unofficial expert. The org chart says one thing. The text messages say another. 3. If a new hire had to do this thing tomorrow, would you have to spend a full afternoon writing it down before they could do it without you? If yes, the writing-it-down is the product. 4. Have you been doing this thing so long you’ve forgotten when you learned it? This is the hardest one to answer honestly, and it’s the one that matters most. The thing that’s automatic for you is the thing somebody else would pay twenty-five dollars to learn.

Three out of four yeses means you’ve got the old skill Claude can amplify. One or two yeses means keep looking — there’s another skill in your week that you’ve discounted, and it’s probably the one your coworkers ask you about the most.

The reason why this test is important is simple. The skills that feel valuable to you are usually the ones you’re still learning. The skills that are actually packageable are the ones that have gone invisible because you do them so often you’ve stopped noticing.

The 3-step pattern: skill, playbook, ship

This is the pattern Edmund ran, written so it transfers to any desk-job skill — not just Chrome extensions.

Step 1 — Pick the old skill, not the cool one

The first move is not to learn AI. The first move is to name the existing skill you’ll point Claude at. Edmund picked Chrome extensions because that was the thing he’d been doing for years at his desk job. You pick whatever the equivalent is for you. Spreadsheets. SOPs. Project templates. HR checklists. Quote-to-order workflows. Process audits.

Pick the skill that has the most callused part of your brain attached to it. The one you do without thinking. Use the four-question test from the section above.

Step 2 — Build YOUR personal Claude playbook on a Saturday

This is the step that separates the working adults who actually ship from the working adults who watch ten Claude videos and never open Claude. Open Claude on a Saturday morning. Spend one weekend. Friday night to Sunday afternoon if you want a deadline.

Name your commands after the steps you take every week at work. If your skill is bookkeeping, you might name one “draft monthly summary.” If your skill is project management, “weekly status report.” If your skill is sales ops, “pipeline cleanup.” Whatever parts you redo over and over — those are the command names.

Build your agents around the heavier work, the parts you keep redoing where each one is a little different. Edmund built eleven of these tuned to his specific Chrome extension workflow. You build yours tuned to your skill.

The compounding move here is that you do this once. One weekend buys you the next year of shipping. Every product you build after that, your playbook is already there. That is what compounds under Edmund’s revenue. It’s not Claude doing the work. It’s HIS Claude playbook doing the work.

Step 3 — Ship on Gumroad or your own site

Once the skill is named and your Claude playbook is in place, the file goes somewhere people can pay you for it. The two answers worth using in 2026:

Gumroad is fastest. About thirty minutes of setup. Upload the file, write the description, pick the price. You own the customer email, which matters when you ship version two.
Your own site if you want control over the experience end-to-end, or if the product is software that needs installation.
The Chrome Web Store if your product is itself a browser extension (this is Edmund’s case).

For digital tools, I wouldn’t use Etsy. Etsy owns the customer, not you. The whole reason you’re building this is to own something — a customer list, a product, a small revenue line you control. Don’t trade ownership for foot traffic on day one.

Still not sure which online business fits you? I built a free 2-minute quiz at finder.platformproof.com that takes you through the decision based on your skills, your time, and what you already know. Most people finish it in under 3 minutes and walk away with a specific next step.

The “but he’s a developer” objection — and why it dies on the math

This is the spot where most working adults watching this freeze. They see Edmund’s Chrome extension and think, “Okay, but he’s a developer. I’m not a developer.” Yeah. Watch this.

Edmund’s old skill happened to be building Chrome extensions. Not because Chrome extensions are special. Because that was his desk-job skill before Claude existed. The shape of the pattern works whether your desk-job skill is spreadsheets, SOPs, project templates, HR onboarding flows, vendor-management processes, or any of the other skills working adults build over a decade and then stop noticing.

Run the numbers one more time. $25 × 120 sales = $3,000 a month. The math doesn’t care if you’re a developer or a working adult with a spreadsheet on a shared drive. The path doesn’t care either:

`Job → Skill → Product → Path`

Same idea every time, whether the skill is code, accounting, or a checklist nobody else can run as cleanly as you can. The proof is one Chrome extension shipper in his apartment, one public Indie Hackers page, and one public GitHub repo anyone can open right now.

Edmund’s honest six-month curve (the part the highlight reels skip)

Here’s the part nobody hyping AI tells you, and it’s the part that matters most if you’re a working adult with a day job and limited weekends.

Edmund did not hit $3,700 a month in his first month. He hit it in month six. The honest curve on his Indie Hackers product page looks like this:

Month 1: roughly $400 in MRR
Month 3: roughly $1,200 in MRR
Month 6: roughly $3,700 in MRR

Six months. Not thirty days. Not overnight. Not a viral launch. A real founder shipping iteration after iteration of one product, with one playbook, on weekends.

This is the part of the math that’s important. Six months from now is going to happen anyway, whether you build the thing or not. Edmund used those six months to build something that pays him every month, and you’ll likely be in the same chair, with the same skill, watching the same videos, unless you decide that the next six months go differently. That’s not motivational filler. That’s just calendar arithmetic.

The one move on Edmund’s GitHub that nobody is talking about

Here is the tactical move you can lift on Monday morning regardless of what your skill is. Edmund didn’t just use Claude. He built his own personal Claude Code playbook and pushed it to GitHub. The repo is named `edmunds-claude-code`. It is public. You can open it right now.

Inside that repo:

Fourteen named slash commands tuned to his exact workflow
Eleven specialized AI agents for the parts he keeps redoing
A README that explains how the whole system fits together

The first time he set it up took him a weekend. Every project after that, those fourteen commands stack up. The agents do the heavier lifting. The playbook is the asset. Claude is the engine. The skill is the fuel.

That is the lift. The advice on most Claude videos boils down to “use better prompts.” That advice keeps you renting Claude’s capabilities one prompt at a time. Edmund’s move is the opposite — he owns a playbook of commands and agents that he reuses on everything he ships. He’s not asking Claude to remember his workflow. The repo remembers his workflow. Claude just runs it.

Saturday morning, two months from now

Picture this. Two months from now. Saturday morning. You wake up before your alarm. You open the Claude repo you spent last weekend building. Fourteen of your own slash commands, six of your own agents, all of them named after the parts of your job you used to dread. You ship the third version of your product before your second coffee. Your spouse walks past, sees you working, asks if you’re checking sports scores again. You don’t even look up. You just say — “no. The score is three thousand dollars in monthly recurring revenue.”

That’s the version of you who built the playbook. The version who didn’t is still scrolling AI-agency videos. The difference is one weekend in May.

What this means for working adults in 2026

If you watched four “make money with Claude” videos this month and walked away each time thinking the AI lane is owned by twenty-three-year-olds in San Francisco, here is the reframe that actually fits the receipt.

You walked in thinking the people quietly winning with Claude were AI agency founders, course sellers, and tech-bros in San Francisco. You walk out knowing the people quietly winning at this are working adults building their own playbooks on Saturday mornings, with skills they already had before Claude existed.

You’re not somebody who watches Claude videos and feels behind anymore. You’re somebody with a job skill, a Claude playbook waiting to be built, and a verified path. Edmund’s been doing this for six months, and he started from the same chair you’re sitting in right now. You’re one weekend away from your own version of his first month.

My goal with Platform Proof is to help one million working adults make their first $3,000 online with the skills they already have — the boring ones. The ones you’ve stopped seeing as skills. The ones your coworkers text you about. The ones that pay $25 a copy when you finally write them down and ship.

Do this today

If you only do one thing after reading this:

Open Edmund’s Indie Hackers product page and his public GitHub repo in two browser tabs.
Verify the numbers. Read the README. Notice that the playbook is the asset.
Take the free Side Hustle Finder quiz to name the skill of yours that Claude should amplify. It takes under five minutes and points at the one your coworkers already text you about.

If you want the deeper playbook for turning your own skill into a Claude-powered product — including the seven-day plan and the pricing-and-positioning math behind Edmund’s $25 price point — that lives in OfferEngine. Single $17 product, no upsell wall. Worth it for the math alone if the math is what got you this far.

When you’re done with the quiz, watch the full video version of this breakdown and grab the one-page Action Notes worksheet so the four-question test and the three-step pattern sit on your desk next to your real-life skill.

Go build the damn thing.

[IMAGE 1: alt=”How to use Claude to build a $3K/month side project — verified Indie Hackers receipt for Easy Folders Chrome extension” | concept: “screenshot of the Indie Hackers product page header for Easy Folders showing $3,700 MRR and $42K total revenue, with a calm-knowing portrait of Alston anchored on the left and a Claude logo bottom-right”]

[IMAGE 2: alt=”The four-question test for spotting a sellable old skill” | concept: “Clean infographic listing the four questions vertically with gold checkboxes on a deep navy background”]

[IMAGE 3: alt=”Edmund Yong six-month MRR curve from a Claude-built Chrome extension” | concept: “Simple line chart showing four data points: $0 launch, $400 month 1, $1,200 month 3, $3,700 month 6”]

[IMAGE 4: alt=”Public GitHub repo edmunds-claude-code with 14 slash commands and 11 specialized AI agents” | concept: “Screenshot-style render of a GitHub repo file tree with folder names for slash-commands and agents visible”]

[IMAGE 5: alt=”Three-step Claude side project pattern — skill, playbook, ship” | concept: “Three-column diagram, each column an icon and one-line label: SKILL (briefcase), PLAYBOOK (notebook), SHIP (paper plane)”]

Sources

Word count: 3,090