Loading page…
Generate a complete PRD + technical docs with CodeGuide first, then hand them to Replit for step-by-step vibe-coding — replicating a full all-in-one agency SaaS with accounts/permissions/Stripe payments without writing a single line of code
The moat isn't the code but the docs-before-code workflow — CodeGuide turns your vision into a structured PRD/tech stack/flowcharts so the AI never free-writes at a blank screen; cloning a proven money-maker is far lower-risk than dreaming up ideas from zero
Fewer bars = easier, cheaper, or faster for an AI-assisted solo builder. Editorial judgments based on the case details.
The video explicitly names the reverse-engineered product: Iman Gadzhi's SaaS “Floe,” an all-in-one platform for agencies, digital businesses, and consultants — client management, invoicing, tasks, CRM, live chat, the “Brain” knowledge base, and visual dashboards — self-reporting over $1.3M/mo. The creator stresses he is “legally building a functionally identical replica,” not an authorized version.
A CodeGuide PRD plus a Replit build gets you a working agency platform in an afternoon; it does not get you the thing that makes Floe work, which is an owner who already has an audience of agency operators. A replica starts at zero there. For all-in-one agency software nobody arrives by search either -- an owner typing "agency management software" lands on HoneyBook, Dubsado, ClickUp or the original. So the realistic cold start is one agency the builder can reach by phone. Run it free for a single shop of five to fifteen people, migrate their client list, open invoices and task board by hand, and let the pain of that migration set the roadmap. After that, agency owners cluster in identifiable places: paid communities and Skool groups for agency owners, the marketing-agency corners of YouTube and X, and larger agencies that would white-label the portal for the smaller shops they already advise. Ten workspaces is the first real milestone, and all ten are sold one conversation at a time.
Taking money is the easy part of this build: a publishable key, a secret key and a prod_ product ID pasted into Replit, and the platform can charge the same day. What the demo leaves open is what to charge for, and that answer comes from the stack the product replaces. An agency already paying for a client portal, an invoicing tool, a CRM and a chat workspace is spending somewhere between $150 and $300 a month across four subscriptions, so a single workspace price in the $79 to $149 range reads as a saving rather than a new cost. Charge the first agency before the product is finished, at a founding rate held for life, because an invoice is what turns a friendly favour into honest feedback -- unpaid pilots go quiet instead of complaining. And offer annual upfront from the first deal. For a solo builder with no runway, twelve months collected at signup is what pays for the hosting, the Stripe fees and the next three months of fixes.
It delivers the most value viewed as an AI build-workflow tutorial: the CodeGuide docs-before-code paradigm genuinely cuts vibe-coding failure rates. But “clone someone's SaaS making millions a month” is clickbait — the truly hard parts are acquisition and compliance, exactly what the video never touches. Fine as a practice project for learning the stack; don't treat it as a money-making blueprint.
Paste into Claude Code / Codex and get a working version of this product end-to-end.
Two production-grade prompts per idea — a full build spec and an SEO growth plan — plus 10 tailored build specs a week. Members only.
Join the Unicorn Club — $5/moCancel anytime, in one clickEvery claim traces back to a listed source — open any of them and check the number yourself.
Data credibility: ⚠️ The $1.3M/mo is the self-reported revenue of the cloned target — Iman Gadzhi's SaaS “Floe” — and has not been independently verified. The video creator merely “legally replicated” a functionally identical copy as a teaching demo, showed no revenue of his own, and the video is essentially a build tutorial funneling viewers to a CodeGuide discount code. This is secondhand reporting of someone else's results. All revenue figures are founder-reported or estimated by the video creator — unaudited.
The role structure the build already implements -- admin, team member, client -- is also the only free distribution this product has. Every agency running on the platform invites its own clients into the portal, and those clients are usually businesses that buy services and sometimes run teams of their own. A branded portal that an agency's customers open weekly is a recurring impression in front of exactly the right buyer, at zero acquisition cost, which is why the client tier deserves the most design attention rather than the least. Two other things compound. Switching cost builds fast once invoices, CRM history and chat threads live in one workspace, so retention climbs with time in the product rather than with features. And an agency that manages other agencies becomes a channel by itself. The honest turning point for a vibe-coded stack is narrower than revenue: it is the first workspace that runs six months of live invoicing without a support emergency.
The build itself is a small rehearsal for the long version. The login loops back on itself, every dashboard component freezes after signup, chat messages send but never render, the View button logs users out, and clients can reach a CRM they should never see -- each one fixed by describing the symptom and running the loop again. That rhythm, symptom in and patch out, is the same one that carries the business through its quiet months, just slower. The trap afterwards is that features are the cheapest thing to produce, so a stalled product gets more modules instead of more users. The numbers worth watching are narrow: how many workspaces sent an invoice this week, how many agencies completed a migration after signing up, and whether any client-tier user logged in twice. A signup that never migrates is a migration problem, not a feature gap. And if the original owns the audience, the replica survives by taking a slice it ignores -- one country's tax and invoicing rules, or one narrow type of agency.
First describe the vision in CodeGuide (with its built-in Cody voice assistant); the AI probes on role permissions (admin/team member/client), the color scheme (purple-to-blue gradient #531FC0→#1989FC), chat capabilities (files/voice/group chat/emoji), and more, then generates the PRD, tech stack, front-end/back-end specs, flowcharts, and a complete implementation plan. Then paste the PRD into Replit and have the AI parse it before coding, choosing PostgreSQL and configuring advanced invoicing (taxes/multi-currency).
The demo is full of faceplants: a login dead-loop, every dashboard component frozen after signup, chat messages sending but never displaying, emoji/file uploads working only intermittently, “Pay Now” fine while the “View” button triggers a logout, and clients able to reach the CRM beyond their permissions. Every issue got fixed by feeding the symptoms back to the AI over and over — showing the real rhythm of vibe-coding is nothing like one-click generation.
The creator never shows any income of his own; the $1.3M is purely the cloned party's self-report, and the video's real engine is affiliate lead-gen for a CodeGuide discount code (Mikey 10% off). The genuinely transferable asset is the PRD-before-code paradigm; but “clone it and profit” ignores acquisition, brand, and legal barriers. Treat it as a technical exercise, not a business blueprint.
That was one of the 50 breakdowns open to everyone. The Unicorn Club opens the other 1,285 at the same depth — plus pain-point filters and 10 build specs a week.
Join the Unicorn Club — $5/moCancel anytime, in one click