For founders who are also the QA team
You shipped it alone.
Don't test it alone.
You're the founder, the engineer, the designer, the support rep, and the on-call rotation. You do not have time to hand-write Playwright suites. Lastest gives solo SaaS builders real E2E + visual tests in 5 minutes. Click through your app once and AI turns it into deterministic tests that run free forever.
Founder Check
- Every solo founder, eventually
- You, staring at your Stripe dashboard
- Also you, every Friday
Why Not Just Click Through It Yourself?
You have four options as a solo founder. Three of them lose you customers.
80% of indie hackers spend over a month shipping their first MVP. The ones who survive the next year spend 6–7 hours per month manually clicking through their own app before releases, because they can't afford downtime and they can't afford a QA contractor. Four ways to handle this, and where each one breaks.
Click through it yourself before every push.
The default. You open the app, sign up, check the paywall, run a test checkout, click around the dashboard, swap browsers, pray. You do this before every release. You forget two flows every time. One of them is always the one that breaks.
- 6–7 hours per month of manual clicking (measured)
- Human memory is not a regression suite
- You skip it the week you're tired - that's the week it breaks
Hire a QA freelancer on Upwork.
$500–$2,000/month for coverage you can't verify and a person you have to onboard every sprint. They test whatever they remember to test. They leave for a better gig. You're back to square one with no tests and a drained runway.
- Burns runway - 2–5% of MRR for a solo doing $20k/mo
- Coverage drifts - no artifact, no git history, no diff
- Knowledge walks out - every contractor churn resets you
- Still can't run in CI on every commit
Write Playwright tests by hand, like the big kids do.
You take a weekend, you sit down, you write 20 tests. They're brittle because you used Tailwind class selectors. They're DOM-only so they pass when the paywall visually collapses. You spend another 6 hours/month maintaining them and eventually delete half.
- Brittle selectors - break on every redesign
- DOM-only - the checkout visually broke, tests pass
- Zero visual regression - Stripe UI update? You won't know until users scream
- You stop writing new ones because maintenance kills you
Use Lastest. Record once. Replay forever.
Click through your signup, paywall, checkout, and dashboard flows once. Lastest captures them as deterministic Playwright code with 7-layer self-healing selectors and a visual baseline. Every push replays the full suite: zero tokens, zero maintenance, zero SaaS bill.
- 5-minute setup - docker-compose up, record a flow, done
- $0 forever - FSL-1.1-ALv2, self-host or use the hosted free tier
- Visual diff catches what DOM tests miss - 3 engines (pixel, SSIM, perceptual)
- Self-healing selectors - 7-layer fallback survives AI refactors
- Works with Stripe test mode, Supabase auth, Shopify preview
- You're the only reviewer - one-click approve/reject on diffs
“The test suite you don't write is the one that refunds your users. You're a one-person company. You need a one-click safety net.”
Built for One-Person Companies
Ship on Friday. Sleep on Saturday.
Four things Lastest does that actually matter when you're the only person who can fix it at 2am.
01 · Record
Click through signup, paywall, and checkout. That's the test.
Open the recorder, run the flow you care about (signup, Stripe test card, dashboard), then hit stop. Lastest generates deterministic Playwright code with 7-layer fallback selectors (data-testid → id → role → aria → text → css → OCR). No selectors to maintain, no docs to read, no weekend gone.
- Captures file uploads, downloads, clipboard, network calls
- Works with Stripe test mode out of the box
- Multi-step screenshots so you see exactly what happened
02 · Generate
Paste your URL. Get a test suite back.
Don't even have 5 minutes? Paste your deployed URL. Lastest crawls it, identifies your auth, pricing, and dashboard flows, and generates tests automatically. Use Claude, OpenAI, OpenRouter, or local Ollama: your keys, your budget, your models.
- Spec-driven from markdown, OpenAPI, or user stories
- Detects framework: Next.js, Remix, Astro, SvelteKit
- AI Test Fixing when something breaks (up to 3 attempts)
03 · Autonomous
The 11-step pipeline that tests your MVP while you sleep.
The Play Agent runs Scout → Planner → Generator → Healer → Reporter. It discovers every route in your app, applies a SaaS template, writes tests, runs them, fixes what it can, and leaves a punch list of what it couldn't. You wake up to a green dashboard or a clear list of three things to look at.
- Auto-discovers routes from Next.js / Remix / Astro source
- SaaS, marketing, and e-commerce templates
- WCAG 2.2 AA accessibility scoring on every build
04 · Catch silent breakage
Did the Stripe redesign just break your paywall?
Lastest captures screenshots on every run and compares them with three diff engines: pixelmatch, SSIM, and Butteraugli. Text-aware masking ignores timestamps and user IDs so you only see real regressions. Review with a slider, approve or reject in one click.
- Auto-mask timestamps, UUIDs, rotating headlines
- AI failure triage: real regression vs flaky vs env issue
- Branch-to-branch compare for pre-deploy review
MCP Server
Your coding agent can test your app now.
Lastest ships with a Model Context Protocol server that exposes ~20 tools to any MCP-compatible agent: Claude Code, Cursor, Windsurf, or whichever one is helping you ship tonight.
When you ask Claude to “add a discount code to the paywall”, it can also run the checkout test, surface the new baseline for your approval, and tell you exactly what it broke, all without you switching tabs.
One install. Zero boilerplate. Your entire test suite becomes a first-class tool in any agentic workflow.
# Install the MCP server
$ npx @lastest/mcp-server
# In Claude Code:
> run the stripe checkout suite
✓ calling lastest.run_tests { area: "checkout" }
✓ 8 tests queued
✓ 7 passed, 1 visual diff flagged
> show me the diff
✓ calling lastest.review_diff
paywall price moved 12px, Stripe
element re-rendered. approve?
Plays nice with your solo stack
Generate once.
Replay forever.
On a $5 VPS.
Solo founders count every dollar. Lastest uses AI only when you create or fix a test. Every subsequent run is pure Playwright: no AI calls, no per-screenshot fees, no CI bills that scale with your test suite. You can run the full suite on every push all month and still fit inside GitHub Actions' free tier.
Self-host or use the free tier
docker-compose up -d on your $5 Hetzner / DigitalOcean box, or sign up at app.lastest.cloud. Your screenshots never leave your infra. Your Stripe test data is yours.
Wire it into your deploy
Vercel preview URL → Lastest GitHub Action → block merge if a flow breaks. Smart Run analyzes git diffs and runs only affected tests, so your CI stays under budget even as you scale.
Bring your own AI (or none)
Claude CLI, OpenRouter, Anthropic, OpenAI, or local Ollama. Test generation is the only AI cost. Replays cost $0. If you're bootstrapping, use Ollama on the same box and your total monthly cost is just the $5 VPS.
Common questions for solo founders
- I'm a solo founder. Why do I even need E2E tests?
- Because you are the only line of defense. There is no QA team, no on-call rotation, no junior engineer to catch the regression you introduced at 11pm. Every broken checkout is a refunded user. Lastest gives you a safety net you can set up in five minutes and maintain in zero: replays are deterministic, free, and run on every push.
- How do I test a Stripe checkout flow end-to-end as a solo founder?
- Use Stripe test mode (card 4242 4242 4242 4242), point Lastest at your app, click through the full signup → paywall → checkout → success flow once, hit stop. Lastest captures the flow as deterministic Playwright with visual diffs and replays it on every push. You get a failing build the day Stripe changes their checkout UI or your paywall copy silently breaks.
- My app is on Supabase. How does auth work in Lastest tests?
- Two options. Easy: record the login flow and let Lastest drive the UI. Fast: use a Playwright setup file that logs in via the Supabase REST API once and reuses the session across tests, roughly 10x faster and way less flaky. The full walkthrough lives at /for/supabase-apps.
- I vibe-coded this on Lovable, v0, or Bolt. Can I still test it?
- Yes. Lastest tests the running application, not your codebase. It does not care whether the HTML was written by you, a framework, or an AI agent. Point it at the deployed URL or a local preview, record a flow, done. Dedicated walkthroughs at /for/lovable-apps, /for/v0-apps, and /for/bolt-apps.
- How much time does Lastest actually cost me each week?
- Setup is roughly 30 minutes for docker-compose up and your first recorded test. Ongoing: a few minutes to approve or reject a visual diff when one fires. That replaces the 6–7 hours per month solo founders typically spend manually clicking through their own app before releases, based on real indie projects.
- Is there a hosted version so I don't have to run Docker?
- Yes. app.lastest.cloud is the hosted option: same product, zero ops. The self-hosted version is free under FSL-1.1-ALv2 and runs on any VPS with docker-compose. Most solo founders start hosted, then move self-hosted when they hit scale or care about data residency.
- I only have one machine and no CI. Can I still use Lastest?
- Yes. Run the suite locally via Docker before every deploy, or wire up the GitHub Actions free tier (2000 min/month private, unlimited public). Lastest's replays are pure Playwright with zero AI calls, so you will not burn your free minutes even on a noisy commit day.
- What if a regression ships anyway? What does recovery look like?
- Every Lastest build stores screenshots and diffs keyed to git SHA. When a user reports a bug, open the dashboard, find the last green build, compare it to HEAD, and see exactly which pixels changed. It is the difference between "bisecting for an hour" and "I see it, it is on the pricing page, revert this commit."
Your first test suite
in five minutes.
No cloud account. No API key. No procurement. No seat limits. No per-screenshot fees. Open source, FSL-1.1-ALv2, $0 forever. Runs on your laptop, your $5 VPS, or the free hosted tier.
★ Star & Install from GitHub →# 1. clone
$ git clone https://github.com/las-team/lastest
# 2. start
$ cd lastest && docker-compose up -d
# 3. record your first checkout test
$ open http://localhost:3000
✓ ship on friday. sleep on saturday.