I'll verify current pricing on the key tools since the budget constraint is a hard one, then build the full package.Before the package: I pulled current pricing on the hosting and database pieces because your "under $50/month" line is a hard constraint, and two of the defaults have a catch that directly affects it. That shows up in the Technical Decisions section. ## 🎯 Project Vision **A mobile-friendly web app that lets independent freelancers and small agencies log billable hours against each client, turn a month's hours into a PDF invoice, collect payment through Stripe, and automatically chase unpaid invoices.** **Purpose.** Replace the spreadsheet-plus-manual-invoice routine that hourly freelancers use today with one place that goes from "I worked 3 hours for Client A" to "invoice sent, paid, and reconciled" without leaving the app. **Users.** Independent freelancers and small agencies who bill clients by the hour. They are not technical. They live on their phones as much as their laptops, so the app must feel native on a small screen. **Core features (v1).** - Log billable time per client (date, hours, optional note), with an hourly rate set per client. - Generate a monthly PDF invoice per client from the logged hours. - Send the invoice and collect payment via Stripe. - Automatically remind clients about invoices that are unpaid past their due date. - Individual accounts with private, secure login. **Success criteria.** A freelancer can sign up, add a client, log hours across a month, generate and send a correct PDF invoice, get paid through Stripe, and have the app chase a late payment on its own — all from a phone, with the running cost staying under $50/month. **Build context.** Built solo by a beginner who is comfortable with no-code tools but new to databases and hosting, using an AI coding assistant to write the code. Target: live within 8 weeks. Stripe integration is non-negotiable. --- ## 🧩 Assumptions and Open Questions **Confirmed from your inputs** - It's a web app (not a native mobile app), and it must work well on mobile. - Features: per-client billable-hour tracking, monthly PDF invoice export, unpaid-invoice reminders. - Users are hourly-billing freelancers and small agencies. - You're a beginner, at ease with no-code, new to databases and hosting. - Budget: under $50/month to run. - Timeline: launch within 8 weeks. - Stripe integration is required. - You're building solo with an AI coding assistant. **My assumptions** (my inferences, not things you told me — please correct any that are wrong) - Stripe is for collecting your clients' payments on the invoices, not for charging freelancers a subscription to use the app. This is the most natural reading of "reminders for unpaid invoices." - Each freelancer gets their own single-login account. "Small agencies" are treated as one account for v1, not multiple team members sharing one workspace. - Default currency is EUR with a single tax line (you're in Spain, so IVA is likely relevant), configurable per user later. - "Works on mobile" means a responsive website, not an app in the App Store. - Hours are entered manually (type them in, or a simple start/stop timer), not pulled automatically from calendars or other tools. - Invoicing is monthly and hours-only for v1 — no expenses or flat-fee line items yet. **Questions to answer next** 1. Is Stripe meant to collect **client payments on invoices**, to charge **freelancers a subscription** for the app, or both? This changes the payment design significantly. 2. Do invoices need **VAT/IVA and configurable tax**, and do you need more than one currency at launch? 3. Do small agencies need **multiple team members** on one account in v1, or is a single login acceptable for the first release? 4. Should invoices support **expenses or flat-fee line items**, or strictly hours × rate? 5. Do you want a **start/stop timer**, or is manual hour entry enough for v1? 6. Do you have (or can you set up) a **business entity and bank account** for the Stripe account that receives payouts? 7. Is there a **business model for the app itself** (free, paid subscription)? If it's paid, that's a second, separate Stripe use case. --- ## 🏗️ Technical Decisions Think of your app as having four parts: the thing users see (the **framework**), the place it lives on the internet (**hosting**), the filing cabinet that remembers everything (the **database**), and the bouncer that checks who's allowed in (**authentication**). Plus payments, which you've already committed to. **Framework — how the app is built** - **Recommended:** Next.js (a popular way to build React web apps). - **Alternatives:** Plain React with Vite; a no-code builder like Bubble. - **Why it fits:** Next.js is the single best-documented web stack for AI coding assistants, so your assistant will make far fewer mistakes and you'll find an answer to almost any error online. It produces responsive, mobile-friendly pages out of the box, and it pairs cleanly with the hosting and database below. A no-code builder would feel more familiar to you at first, but it fights you the moment you need custom Stripe and PDF logic — which is most of this app. - **Flag:** Reversible early, painful late. Switching frameworks after you've built several features means a near-total rewrite, so this is a decision to make once and keep. **Hosting — where the app lives** - **Recommended:** Vercel. - **Alternatives:** Netlify; Cloudflare Pages; Railway. - **Why it fits:** Vercel is made by the team behind Next.js, so deploying is close to one click, and it includes scheduled jobs (needed for your reminders) at no extra charge. For a beginner new to hosting, this removes almost all the scary parts. - **⚠️ Flag (cost + rules):** Vercel's free "Hobby" tier is restricted to non-commercial, personal use only, and all commercial usage requires the paid Pro plan. The moment you turn on Stripe to take money, you're a commercial site. Pro is $20/user/month and does allow commercial use. Budget for this from launch day. You can still build and test for free during the 8 weeks; you only need Pro when real money starts flowing. **Database — the app's memory** - **Recommended:** Supabase (a managed Postgres database with auth and file storage bundled in). - **Alternatives:** Firebase; Neon (database) paired with a separate auth service. - **Why it fits:** Supabase gives you the database, the login system, *and* a place to store the PDF files in one product with one dashboard — which is exactly the "I don't want to wire five services together" setup a beginner needs. It uses standard SQL (Postgres), which keeps your data portable if you ever leave. - **⚠️ Flag (cost + reversibility):** The free tier looks generous but free projects pause automatically after 7 days of inactivity and have no automated backups — both unacceptable for a live app holding people's invoices. The paid Pro plan is $25/month and adds backups and keeps the project always-on. Reversibility is medium: your raw data is portable, but Supabase's login and security features create some lock-in, so moving off it later is real work, not a copy-paste. > **Running-cost reality check.** Vercel Pro ($20) + Supabase Pro ($25) = **$45/month**, before a cheap domain (~$1/month). That fits under $50, but with almost no headroom. Any extra paid add-on will break the budget. Practical plan: build on the free tiers (≈$0) for the full 8 weeks, then flip both to Pro at launch. **Authentication — the login system** - **Recommended:** Supabase Auth (comes with the database above). - **Alternatives:** Clerk (very polished, beginner-friendly); Auth0. - **Why it fits:** Since it's already part of Supabase, you avoid stitching a separate login service into your app — one fewer thing to configure and pay for. Email-and-password plus "sign in with Google" covers your users. - **Flag:** Medium reversibility. Once real users have accounts, changing your auth provider means migrating everyone's login records — doable, but a chore. Fine to accept given it's bundled and free. **Payments & invoicing — the part worth thinking hardest about** - **Recommended:** Stripe, and specifically lean on **Stripe Invoicing** rather than building invoices from scratch. - **Alternatives:** Build invoicing yourself and use Stripe only as a payment button (Stripe Checkout / Payment Links); PayPal (ruled out — you specified Stripe). - **Why it fits:** This is the biggest scope-saver in the whole plan. Stripe Invoicing can create an invoice, generate the **PDF**, email it to the client, **collect the payment**, and send **automatic reminders for unpaid invoices** — which is three of your four core features handled by a service you're already required to use. That means the only truly custom thing you must build is the *hour-tracking* layer; then you hand a list of line items to Stripe and let it do the rest. For a solo beginner on an 8-week clock, this is the difference between shipping and not. - **⚠️ Flag (cost + reversibility):** Stripe has no monthly fee but takes a per-transaction cut (for European cards, roughly 1.5% + €0.25; confirm current rates for your account and country). That's variable cost that scales with usage, not a fixed bill — good for a tight budget. Reversibility: you always own your hour data, so if you ever left Stripe Invoicing you'd rebuild the invoice/PDF/reminder layer but lose nothing essential. Low lock-in risk given you own the valuable part. --- ## 🔌 Recommended Tools and Integrations - **Stripe (+ Stripe Invoicing)** — Handles payment collection, invoice PDFs, and unpaid reminders in one service, covering most of your feature list. *Setup:* create a Stripe account, complete business verification for payouts, then connect it to your app using Stripe's official Next.js guide and test everything in Stripe's free "test mode" before going live. - **Resend** — Sends the emails your app controls (welcome emails, and any reminders you handle yourself rather than via Stripe). It's the most beginner-friendly email service and has a free tier that's plenty for launch. *Setup:* sign up, verify your sending domain (a few DNS records), drop the API key into your app's settings. - **Vercel Cron** — The built-in scheduler that wakes up daily to check for overdue invoices and trigger reminders. *Setup:* it's included with Vercel — you add a small config file describing when to run; no separate account. - **@react-pdf/renderer** — A library that builds invoice PDFs in code, useful if you ever need a custom-designed invoice beyond Stripe's default template. *Setup:* your AI assistant installs it as a code package; no account or cost. Treat this as optional — try Stripe's built-in PDF first. - **Sentry (free tier)** — Tells you when the app breaks for a user, so you're not flying blind after launch. *Setup:* create an account, add the setup snippet your assistant generates. Optional, and the free tier keeps you within budget. --- ## 🪜 Staged Build Plan Each stage is small enough to test on its own, and the order is deliberate: your two weakest areas (hosting and databases) get de-risked *first*, before you've built anything you'd hate to lose, and the trickiest integration (Stripe) comes only after the core value exists. **Stage 0 — Skeleton on the internet** - **Goal:** Get a blank Next.js app deployed to Vercel and connected to an empty Supabase database. - **Done when:** A near-empty page is live at a real URL, loads correctly on your phone, and can read one test row from the database. **Stage 1 — Login** - **Goal:** Add sign-up and log-in with Supabase Auth. - **Done when:** You can create an account, log in, log out, and a logged-out visitor can't see private pages. **Stage 2 — Clients and hours (the core)** - **Goal:** Let a logged-in user add clients (with an hourly rate) and log billable hours against them. - **Done when:** A user can add a client, record several time entries, and see a correct running total of hours and amount owed per client. **Stage 3 — Invoice generation** - **Goal:** Turn one client's hours for a chosen month into an invoice. - **Done when:** Clicking "create invoice" produces a correct PDF (via Stripe Invoicing) showing the right line items, totals, and dates. **Stage 4 — Get paid (Stripe)** - **Goal:** Send the invoice and collect payment through Stripe, in test mode first. - **Done when:** A test invoice can be paid with a Stripe test card, and the invoice's status flips to "paid" in your app. **Stage 5 — Chase late payers** - **Goal:** Automatically remind clients about invoices unpaid past their due date. - **Done when:** An overdue test invoice triggers a reminder (via Stripe's automatic reminders and/or a Resend email on your daily Vercel Cron job). **Stage 6 — Polish and launch** - **Goal:** Mobile QA, error messages, empty-states, then go live and switch to paid tiers. - **Done when:** The full flow — sign up → log hours → invoice → get paid → reminder — works cleanly on a phone, and a real freelancer can use it. **Realistic timeline.** Across 8 weeks, expect roughly: Stages 0–1 in week 1; Stage 2 in weeks 2–3 (it's the heart of the app); Stage 3 in week 4; Stage 4 in weeks 5–6 (Stripe always eats more time than you expect, especially verification and testing); Stage 5 in week 7; Stage 6 in week 8. It's tight but achievable *specifically because* you're leaning on Stripe Invoicing for the PDF, payment, and reminder work. If you tried to build all of that by hand, 8 weeks would not be enough. **Costs to watch.** ~$0/month while building on free tiers, then ~$45/month at launch (Vercel Pro $20 + Supabase Pro $25 + domain), plus Stripe's per-transaction fee. Adding any paid add-on pushes you over $50 — so keep Sentry and email on their free tiers for now. **Scope creep to resist.** The tempting extras that will blow your timeline: multi-currency and tax rules, multi-user agency accounts, expense and flat-fee line items, recurring/retainer invoices, a client-facing portal, and an automatic timer that syncs with calendars. Every one is a fair v2 idea; none belongs in the 8-week v1. Write them on a "later" list and move on. --- Two things would sharpen this a lot: your answer to **Question 1** (is Stripe for collecting client payments, charging freelancers, or both?) and **Question 3** (do agencies need multiple team members in v1?). Both change the data model, so they're cheapest to settle now. Want me to turn this package into a Markdown or Word file you can hand to a fresh AI session?