Four AI agents, 42 days, six live products, one sale. The most useful page is the list of what still needs a person.
Walker Brown let four coding agents build and sell their own products, then published everything: 4,376 commits, six live products, one $19 sale that netted $18.09 (self-reported, from the project's Stripe record), and a running list of the 20 steps only a person could do. That list points to a business.
Published Oct 21, 2026
The Wow
What someone built
Since Sept 10, 2026, Bananafest Destiny's coding agents have each picked, built and tried to sell a web app with a real domain and Stripe account, under rules they can't edit. As of Oct 21 the live numbers page lists 9 apps attempted, 1 killed, 6 live and $18.09 earned from strangers after fees. Brown says he pays for a max Claude subscription and averages about an hour a day.
- Original source
- Bananafest Destiny: the numbers, live
- License
- Public write-ups; no reuse licence stated. Link and credit only
- To use it
- Nothing to install. Read the numbers page and the 'what still needs a person' list.
- A business you could build on it
- Pre-submission marketplace audits for agent-built apps
Why it's remarkable
It's the most candid public answer to "do AI agents make money on their own?", and it measures exactly where a person is still needed. One agent audited itself against Shopify's 173 listed app requirements and found 15 of the 45 that applied were human-only.
The Opportunity
The business it reveals
Agent-built products stall on small human steps: a two-minute screencast, a dashboard login, a billing plan, a marketplace submission. A fixed-fee pre-submission audit, where the agent runs the checklist and a person handles the human-only items, gets those products unstuck.
Who would pay
Solo founders and small studios submitting agent-built apps to the Shopify App Store or similar marketplaces. The project's own log shows 6 human steps still waiting, the longest for 20 days.
The Flip
3 ways you could turn this into a business
Pre-submission marketplace audits for agent-built apps
- Who pays
- Solo founders and small studios submitting apps to the Shopify App Store or similar marketplaces
- How it would earn
- A fixed fee per audit, with an optional re-check after fixes (our hypothesis; price untested)
- Why it might work
- The project's agent showed the checklist work can be automated, and its own count shows a third of the applicable items still need a person.
- What already exists
- Shopify Partner agencies build and submit apps for clients; Shopify's own review gives feedback after you submit. We found no common fixed-fee pre-check aimed at agent-built apps, but haven't searched exhaustively.
A human-steps desk for agent-run products
- Who pays
- Founders running several agent-built products at once
- How it would earn
- A monthly retainer to clear a set number of human steps within two days (our hypothesis)
- Why it might work
- The project's log shows most human steps take minutes but wait for days, which is a turnaround problem a desk can fix.
- What already exists
- General virtual-assistant services and screencast gigs on freelance marketplaces; none we saw are organized around an agent's queue.
Honest live scoreboards for build-in-public founders
- Who pays
- Indie founders who publish their progress
- How it would earn
- A setup fee for a page that counts revenue, costs and hours straight from their own records (our hypothesis)
- Why it might work
- The project's numbers page earns attention because every figure traces to a record, which readers trust more than screenshots.
- What already exists
- Open-revenue badges from payment and analytics tools exist; few show costs, hours and human steps together.
Why we'd try "Pre-submission marketplace audits for agent-built apps" first: Shopify publishes its requirements as a list, an agent can do most of the checking, and the human-only items are the clear value a person adds.
Build It
How you could build it
A pre-submission audit: requirement-by-requirement status, evidence, fixes, and the human-only items done or scheduled.
First step: Open the Shopify App Store requirements list, have your agent sort every item into 'agent can check', 'needs a person' and 'not applicable' for one app you know, and time how long the person-only items take.
Agents and tools
- Your coding agent
- The marketplace's published requirements
- A screen recorder
- A shared checklist document
Setup
- Copy the marketplace's requirement list into a checklist with a source link per item.
- Audit one public app as a sample report.
- Write a short access guide for collaborator access the platform allows.
- Set a fixed scope and turnaround.
What the agent does
- Checks each requirement against the app and its listing
- Drafts fixes for items it can verify
- Flags the human-only items
- Re-checks after fixes
What you do
- Records screencasts and other hands-on items
- Reviews every finding before delivery
- Never handles identity, tax or payment verification
- Sells and presents the audit
Your first customer: A founder you see preparing a first Shopify app submission, offered the audit as a fixed-fee pilot.
Test demand first: Offer the audit to three founders preparing a submission. Success: one pays for the pilot or reports a blocker it caught. Failure: none take it, which suggests marketplace review is not where they feel the pain.
Your starter instructions
Paste this into the AI agent you use (ChatGPT, Claude, Gemini, Grok Bot or another) as your first message. Copying it starts work in your own agent; it doesn't create or deploy a business by itself.
Running it
What stays human: According to the site: the owner pays for things, answers the agents' questions, posts in communities, records screencasts and clears the steps listed on the 'what still needs a person' page. In the service we propose, a person does the hands-on requirements and the final review.
- Startup cost (estimate)
- A max Claude subscription, per the owner (According to the project's numbers page, the owner says he runs it on a max Claude subscription and publishes no total cost. Our audit service's cost is not measured)
- Running cost
- The project publishes no running total. The owner's time is about an hour a day by his own estimate.
- How much the agent runs
- Agents run one core part; people run the rest.
Pricing your own version? Use the free startup-cost worksheet.
The evidence
What the source shows: The project's live numbers page, its "what still needs a person" page and its DEV posts. The figures are its own and change hourly, so we link the live page.
What's ours: The pre-submission audit is our application.
Worth planning around
- Doing account steps for someone else can break platform terms; use only access the platform lets a founder grant a team member, and never pass identity checks for anyone.
- The market for agent-built marketplace apps is young.
Sources
- Bananafest Destiny: the numbers, live (revenue, costs and hours as of Oct 21, 2026)
- Bananafest Destiny: what still needs a person
- Bananafest Destiny: meet the team
- Walker Brown on DEV: Week three — 989 commits, 30 impressions, zero clicks, $0
- Shopify App Store requirements
Sources reopened Oct 21, 2026. Spot a mistake? Tell us and we'll correct it in public.
Keep reading
- Services to sell to people building apps with AI (guide)
- How much of the work do AI agents actually run? (guide)
- Can AI agents make money on their own? (guide)
- He still can't write code. His five-person ad agency now runs on about 270 tools he built with Claude. Every owner-run firm needs its first five
- A marketer who doesn't code built an elk-call coaching app with Claude Code and says it hit a $650K run rate in 110 days. Every skill learned by sound is the same opening
- A free n8n workflow flags fake supplier invoices before anyone pays them — and Germany's e-invoice switch makes it a service bookkeepers can sell
- Marc Lou's agent chased a broken-shoe refund through three support desks and got his 850 AED back. Small stores now need a claims desk that can answer agents
- Every discovery and launch plan