Idea to funding — a roadmap with a stack you own

From 'I have an idea' to 'a fund wired the money' — the path with a tech stack and a business stack you control end-to-end. No vendor on the critical path of your company.

This is the meta-resource. Every other doc in this library — auth, hosting, email, databases, compliance, the launch sequence, the post-launch playbook — is a single layer in a stack. This page is the map that puts those layers in order, attaches dollar figures and timeframes to each, and tells you when you're allowed to move on. If you read one resource here before you start, read this one.

The premise — own the stack

A vibe coder who finishes this guide ends up with three things: working infrastructure, a small but real set of customers, and a company they actually control. The reason that combination matters is that founders without it tend to discover, somewhere around month nine, that they don't really own the business they thought they were building.

There are three ways people get this wrong, and we'll name them.

The no-code-only path ships fast on Bubble, Softr, or a Glide front-end against an Airtable backend. It works until it doesn't. The platform changes pricing, deprecates a feature, throttles your workflow runs, or simply gets acquired and pivots. Your data is in their schema, your business logic is in their visual editor, and your migration cost is the entire product. We've watched founders with $40K MRR get held hostage by a 4x pricing change because the export was 800 hours of re-implementation.

The off-the-shelf-everything path is the modern equivalent: Clerk for auth, Vercel for hosting, Stripe for payments, Resend for email, Supabase for the database, OpenAI for inference, PostHog for analytics, Linear for the roadmap, Notion for the docs. Each of those is a great service. The composite is a company that has no leverage in any vendor negotiation and no fallback when one of them has a bad afternoon. You don't own auth, hosting, the database, or the inference layer; you own the glue. Glue isn't equity.

The build-everything-in-house path is the overcorrection. Custom auth, self-hosted Postgres on a bare metal box, an in-house mailer with SPF/DKIM you tuned by hand, a custom Stripe replacement because "fees are theft." This founder has a beautiful stack and three customers because they've been at it for fourteen months.

The middle path — the one we teach — is to pick the right level of abstraction at each layer, own the data, own the auth, and keep every vendor swappable. You use Auth.js because it puts the user table in your database (see auth). You use a hosting provider you can leave in an afternoon. You write your business logic in TypeScript that runs anywhere Node runs. The vendors that remain (Stripe, Resend, OpenAI) are commodities with real competitors, and your code touches them through a thin adapter layer. When one of them has a bad afternoon, you swap it. This is the difference between a company you own and a company you're renting.

The 7-stage roadmap

Here is the whole arc, with realistic durations. A founder using this guide and the configurator should expect to see all seven stages inside a calendar year.

Stages 1 through 4 are linear; you do them in order, and each one has a clear gate. Stages 5 and 6 overlap with each other and with the tail of stage 4. Stage 7 is a fork, not a step — there are three serious paths and a fourth path called "stay private," all valid.

The discipline is moving forward only when the stage is genuinely done. Most founders skip stage 2 and pay for it in stage 6.

Stage-by-stage: what counts as "done"

A stage is done when the exit criterion is on paper, not in your head. If you can't show it to a friend, it's not done.

• Stage 1 done = a positioning statement on paper. One sentence: "We help [specific person] do [specific thing] without [the current pain]." If you can't fill those three slots without hand-waving, you don't have an idea, you have an interest.
• Stage 2 done = a vertical-compliance map, a competitive table with at least 5 rows, and a written go/no-go decision with a date on it. See pre-build-due-diligence. If your vertical is healthcare, fintech, or anything touching K-12, this stage is half a day longer and is non-optional.
• Stage 3 done = MVP at a public URL with all non-negotiables shipped. Auth works. Payments work if you're charging at launch. Email deliverability is set up (SPF, DKIM, DMARC; see email). The TOS and privacy policy are real documents. The site loads on a phone. There's a /pricing page.
• Stage 4 done = T+0 executed. Show HN posted, Product Hunt scheduled, the warm list emailed, the founder is on a couch with a beer and a spreadsheet logging incoming signups. See sub-skill 18.
• Stage 5 done = ten interviews logged. Not ten signups — ten conversations, transcribed, with a list of the three things every interviewee said.
• Stage 6 done = PMF signal. This is the fuzziest gate; we'll define it in section 6 below. The short version: retention curve flattening, MoM revenue growth ≥ 15%, and at least one customer who would be visibly upset if you turned the product off.
• Stage 7 done = a wire hits, or you've explicitly chosen not to raise. Both are wins.

Time + money budget

Real numbers. These are not aspirational.

The MVP cash range is mostly OpenAI tokens, a domain, and a year of hosting. The iteration phase grows because you're now paying for production infrastructure, monitoring (see cost-and-observability), and probably a Stripe account that's actually moving money. Funding application costs are legal review of your incorporation and cap table, pitch deck design (skip this if you can use Figma), and travel to two or three meetings.

The whole arc, idea to seed wire: realistically $5K-20K of founder cash and 3-9 months of life. Compare to the traditional path — incorporate with a $5K lawyer bill, go heads-down for a year on ramen, hire an agency to build the MVP for $80-150K, then raise — which lands at $50K-200K easily and 12-18 months. This guide is an order of magnitude cheaper, and the money you don't spend stays as runway.

Funding paths — pick one based on signal

Once you have PMF signal, the path branches. Three serious options, and we'll be honest about which fits when.

Bootstrap / indie. Keep 100% of the equity, grow with revenue, hire only when payroll is comfortably covered by MRR. Fits when LTV > CAC × 3 and the math says revenue can fund itself by month 6. The community is on Indie Hackers; the eventual exit, if you want one, is a strategic acquirer or a marketplace like Acquire.com (the rebrand of MicroAcquire). Most vibe-coded SaaS in the $1K-30K MRR range belongs here.

Angel + small accelerator. $50K-500K total, usually 5-10% equity, gets you 12-24 months of runway plus a network. Fits when you have signal but want capital and connections more than you want full ownership. Y Combinator, Techstars, On Deck, regional accelerators, AI-focused programs like Conviction's Embed — see accelerators for the comparison and the application timing.

Pre-seed / seed institutional. $1M-5M, 15-25% equity, with the explicit purpose of hiring a small team and pushing to the next milestone in 12-18 months. Fits when the market opportunity is large enough that ownership matters less than speed, or when the product needs a larger team to be defensible (anything regulated, anything with hardware, anything where the moat is sales motion).

The wrong path for your situation will eat months of your life. Pick deliberately.

What you need at each funding stage

Concrete asks. Show up with less and you'll waste meetings.

Angel-ready:

• Working product at a public URL
• 100+ users (free is fine if you're pre-revenue; better if some are paying)
• A 1-page narrative — problem, solution, who it's for, what's working
• A personal warm intro to angels in your space (cold pitches to angels rarely work)

Accelerator-ready:

• Working product
• Some metrics: weekly active users, MRR if applicable, week-4 retention
• A clear 6-month plan for what the accelerator capital unlocks
• The application essay — YC's seven questions are the canonical version; most other accelerators ask a subset

Seed-ready:

• Clear PMF signal: retention curve flattening (week-4 retention not still falling off a cliff), revenue growth ≥ 15% MoM at MVP scale, NPS or equivalent qualitative signal
• 5-15 months of MRR data for B2B SaaS, or DAU/MAU history for consumer
• A polished pitch deck (10-12 slides, narrative-led, not a wall of charts)
• A financial model that's honest about assumptions (3-year, monthly granularity for year 1)
• An investor data room with incorporation docs, cap table, customer contracts, and the metrics dashboard live (sub-skill 18 + [pre-build-due-diligence] cover the structure)

The "I'm not raising" path

Worth a section, because the funding-industrial complex will try to talk you out of it.

A vibe-coded MVP runs on $200/month of infrastructure once you stop iterating on it daily. At that cost basis:

• $1K MRR is profitable. Not a salary, but profitable.
• $10K MRR is a real business. A founder with $10K MRR and one part-time contractor is doing better than 80% of seed-funded companies in year two.
• $30K MRR is a $360K/year company you wholly own. That's a top-decile US household income with no boss, no board, and no liquidation preferences ahead of your common stock.
• $100K MRR ($1.2M ARR) is a serious business that you can sell on Acquire.com for a 3-5x multiple, or keep running indefinitely.

Many founders' best move is to stay private and grow at their own pace. The optionality is enormous: at any point you can decide to raise, and the leverage you have when you're already profitable is dramatically higher than the leverage you have when you need the money to make payroll next month. Don't apologize for not raising. The default cultural script that says raising is the success state is a script written by people who make money on raises.

What the agent does at each stage

A quick map of where the agent (Claude Code, Codex, or whatever you're using as your coding partner with this skill set installed) carries the load.

• Stages 1-3: heavy agent involvement. Sub-skills 01-15 are designed to be executed inside the agent loop — the agent walks you through audience definition, runs the due diligence checks, scaffolds the project, wires auth, sets up the database, ships the deploy.
• Stage 4: collaborative. The agent assembles the launch playbook (the Show HN draft, the Product Hunt copy, the email to the warm list), and the founder executes the human parts (posting, replying, reading the room).
• Stage 5: post-build. The configurator's Notes tab is where you log Discovery insights from user calls; the Discovery view aggregates patterns across interviews. The agent helps you spot themes you missed.
• Stage 6: iterative. The agent helps weekly with code changes, feedback synthesis, and the kill list (which features to remove, not add). See testing for the test discipline that lets you ship weekly without breaking things.
• Stage 7: mostly the founder and their network. The agent helps with deck prep, financial model formulas, and the data room file structure. Investors talk to humans.

The compounding effect

Owning the stack has a compounding payoff that doesn't show up until product two or three.

The next product is faster because you've built the muscles. The agent becomes a more effective collaborator over time because there's more context in PROJECT.md and STATE.yaml for it to draw on — your architecture decisions, your taste, your prior mistakes. The cost per MVP keeps dropping because the tooling gets cheaper and the model gets better. The founder accumulates engineering judgment they didn't have at the start of product one — what to abstract, what to leave hardcoded, when to use a library and when to write fifteen lines.

By the second or third product, a founder is shipping in days what used to take months. And the company they're building has the option to stay independent or to raise on terms that reflect a track record, not a pitch.

Where to start

One concrete first action. Pick whichever fits your moment:

1. Install the configurator (see Downloads), open the Mode tab, answer the three questions, and let it pick a mode for you. Then start sub-skill 01.
2. Or, just open Claude Code or Codex and say: "I'm using the Vibe Coder's Guide to MVP skills. Here's my idea: [your idea in two sentences]. Walk me through sub-skill 01." The agent will take it from there.

The roadmap is real. The dollar figures are real. The founders doing this are real. The only thing missing is your stage 1.