I Built and Launched AuditZap

I launched AuditZap today. It’s a website audit tool that runs 40 checks across SEO, performance, and site health, and uses AI to generate fix instructions specific to your CMS. WordPress, Shopify, Webflow, whatever you’re running.

The idea came from client work. Every audit tool I used did the same thing: here are your 47 issues. Good luck. The gap was always the next step - which issues actually matter, and what do I do about them? That’s what AuditZap tries to solve.

How it works

You enter a URL. The app crawls your site (up to 100 pages on Pro), runs the checks, and batches every failed result into a single Claude API call. One prompt, one response, all the fix instructions. That single-call approach keeps the AI cost to about $0.03 per audit.

The checks themselves are pure functions - no side effects, no database calls. Take crawl data in, return a structured result. That made testing straightforward: the suite is roughly 2,975 tests and runs with no Docker, no database, no Redis.

That entry-a-URL step is also the whole of the embeddable widget, which is what you’re looking at here:

If you’d rather run one check than a whole audit, the same crawler powers the free broken link checker, which needs no signup at all.

The stack

Next.js 16, TypeScript strict, Drizzle + Postgres, BullMQ + Redis, Puppeteer for crawling, Claude for AI, Lemon Squeezy for payments. Self-hosted on a single Hetzner box at roughly $20/month, running both production and staging.

I considered Vercel but the combination of Puppeteer (needs a real browser), BullMQ (needs persistent Redis), and background workers made self-hosting the better fit. GitHub Actions handles auto-deploys on push.

Free tools

I built 20 free tool pages, among them broken link checker, meta descriptions, Core Web Vitals, alt text, Open Graph, structured data, HTTPS & mixed content, redirect chains, and image optimisation. No signup, no database writes, results in a couple of seconds. Each one targets a long-tail SEO query. Free marketing that also gives people a reason to come back.

What I’d do differently

Ship earlier. I spent time on things that didn’t need to be perfect for launch - PDF report styling, email template polish, edge cases that affect 1% of audits. The checks and the AI instructions are the product. Everything else is secondary.

Set up monitoring from day one. I added Sentry late and spent a morning retrofitting it into the worker, the crons, the API routes. If I’d done it from the start it would have taken a fraction of the time.

Don’t underestimate auth. Better Auth is good, but email verification, OAuth, two-factor, password reset, session management - it’s a lot of surface area. Every auth feature touches multiple pages and multiple API routes. Budget more time for it than you think.

What’s next

Per-page affected URLs in audit results, so you can see exactly which pages have which issues. Password change in settings. And whatever people ask for - the whole point of launching is to find out what matters to users that I haven’t thought of.

If you want to try it, the free tools work without signing up: auditzap.io/tools.

If you want to support the launch, we’re on Product Hunt today.