The conventional wisdom is that you need money to build a SaaS — hosting, infrastructure, a developer or two. That's not true in 2026. The free tiers on modern platforms are genuinely generous, and AI tools mean a solo developer can build and ship what previously required a small team.
I've built several SaaS apps with minimal budget. Here's the exact approach.
The Free Stack That Actually Works
Frontend: Next.js on Vercel
Vercel's free Hobby tier is extraordinary value. You get:
Unlimited deployments from GitHub
Custom domain support (you pay for the domain, ~£10/year)
Serverless functions (great for API routes)
Edge caching globally
100GB bandwidth/month
Next.js is the right framework for SaaS in 2026 — server-side rendering, API routes, easy auth integration, and Vercel deploys it natively. There's no better combination for a solo founder.
Limitation: Hobby tier isn't for commercial projects with revenue. Once you're making money, move to the Pro tier (£17/month). But for building and validating? Free is fine.
Database: Supabase or PlanetScale Free Tiers
Supabase gives you a full PostgreSQL database, authentication (email, Google, GitHub OAuth), storage for files, and real-time subscriptions — all free up to 500MB. That's enough to run a real product with hundreds of users.
It also includes a REST and GraphQL API generated automatically from your database schema. You can go from idea to working data layer in an hour.
PlanetScale (MySQL-compatible, serverless) has a similarly generous free tier and is particularly good if your data model is simple and you want fast edge queries.
I use Supabase. The built-in auth alone saves a week of development.
Authentication: Supabase Auth or Clerk
If you're on Supabase, use its built-in auth. Otherwise, Clerk is free up to 10,000 monthly active users — which means you can grow significantly before paying anything.
Clerk handles email/password, Google OAuth, magic links, and multi-factor authentication. It takes about 30 minutes to add to a Next.js project.
Payments: Stripe
Stripe has no monthly fee. You pay per transaction: 1.4% + 20p for European cards (a bit more for international). For a SaaS charging £10-30/month, that's roughly 30-65p per subscription. Completely manageable from your first sale.
Set up Stripe with webhooks to handle subscription events (payment failed, cancelled, upgraded). Supabase can listen to those webhooks and update your user database automatically.
Email: Resend Free Tier
Resend gives you 3,000 emails/month and 100/day free. For a small SaaS, this covers welcome emails, password resets, and basic notifications. When you outgrow it, the first paid tier is $20/month for 50,000 emails.
Building It: Use AI Tools Heavily
This is where 2026 is genuinely different. A solo developer with AI tools can build what previously required a team.
My actual workflow:
1. Define the core feature set clearly — write it out as user stories
2. Set up the boilerplate (Next.js + Supabase + Clerk) with AI assistance from Cursor
3. Build features one at a time, using Claude to write complex logic and Cursor to integrate it
4. Use AI to write the copy, landing page content, and documentation
For my Linktree alternative (The Link Spot), I used Floot.com to generate the initial build from a detailed prompt. It created auth, dashboard, Stripe integration, and the public-facing pages in one session. I then refined it, fixed issues, and iterated. Total out-of-pocket cost for the initial build: under £30.
The Build Process (Realistic Timeline)
Week 1: Core product
Set up accounts: Vercel, Supabase, Stripe, Resend, domain
Build the MVP: one core feature, auth, basic dashboard
Deploy to Vercel, test end-to-end
Week 2: Polish and payments
Add Stripe subscription integration
Build the pricing/upgrade flow
Write the landing page
Fix the things that broke during testing
Week 3: Launch prep
Write docs or a simple FAQ
Set up basic analytics (Vercel Analytics is free)
Submit to Product Hunt, post in relevant communities
Tell people about it
That's it. Three weeks to a live, paying-capable SaaS product. Not perfect, but real.
What "Free" Actually Costs You
Be honest with yourself about the non-financial costs:
Time. Building a SaaS solo takes meaningful time even with AI assistance. Evenings and weekends for weeks. Make sure the idea is worth it before you start.
Domain. About £10-15/year for a .com or £5-10/year for a .co.uk. Not negotiable if you want to look professional.
Your own skills. The free stack assumes you can code in JavaScript/TypeScript. If you can't, you need to either learn or use a no-code builder like Floot, Bubble, or WeWeb — which have their own free tiers but different limitations.
Ideas That Work Well With This Stack
The best SaaS ideas for solo founders on a budget are:
Tools that solve a specific pain point for a niche audience (not "for everyone")
Things you would pay for yourself
Projects where the value is obvious enough that explaining it takes one sentence
Ideas where you can reach early users through communities you're already in
Don't build another Notion competitor or Slack clone. Build the tool your industry needs that doesn't exist yet.
After Launch
When you start getting paid users, reinvest the revenue:
1. Move to Vercel Pro (when you need it, not before)
2. Upgrade Supabase if storage/bandwidth is a concern
3. Add proper error monitoring (Sentry has a generous free tier)
4. Hire a part-time designer if the product looks rough
The goal is to get to revenue before spending money. Everything in this stack supports that.
Building in public is a great way to grow early on. Document the process, share what you're learning, and build an audience as you build the product. That's part of what I do on the @PromptToCode YouTube channel — building real products with AI, showing the process honestly. Worth checking out if you want to see this in action.