Live operations room
Watch the agents inspect your system page by page, and ask the whole crew or one agent in a single chat.
Now open · inspect your site free
Catch the flaw before your users do.
Check your website’s critical journeys and Arabic experience in real browsers. Choose from 17 AI specialties and get a clear report with evidence and suggested fixes.
No card · results in minutes
✓Your place is reserved
#1
We sent you a welcome e-mail. Every friend who joins with your link moves you up 5 places.
17 specialistssecurity, UI, performance, access…
Real browsersdesktop, phone and tablet
Minutes, not weeksfrom address to report
Suggested fixesinstructions for your editor
Capabilities
One run. The whole quality bar.
Everything a careful QA team would check before a release — done by a crew that never gets tired.
Code & security
SQL injection, XSS, leaked secrets, headers and exposed files.
UI in real browsers
Chrome, Firefox, Safari and Edge — with right-to-left and accessibility.
Simulated users
Personas from your markets try the journeys and tell you where they got stuck.
Ready-to-paste fixes
A fix prompt for every issue — for Cursor, Claude Code or Copilot.
Inspects safely
Isolated agents, encrypted credentials, and signed-in scans only on a verified domain.
The math
Every bug that reaches your users costs more than you think.
Industry numbers first, then yours.
69%
of developers lose 8+ hours every week to friction and rework
Atlassian · State of Developer Experience 2024252
days, on average, a security flaw stays open
Veracode · State of Software Security 2025Your team
You save every month—
Debugging hours won back—
Return on the plan—
Estimated savings = recovered hours × hourly cost. Return = savings ÷ (plan + extra inspections), before tax.
View the calculation
Community
Your users, before launch.
AI personas try tasks in real browser sessions and identify possible friction. Reports separate the executed sample from the extrapolated population.
Synthetic feedback helps you plan improvements. Validate human behavior with user research, and capacity with a separate load test.
A real betaWeeks
With PreetestMinutes
Sample AI-agent feedback from demo shops — these examples are not customer reviews or reports about Preetest.
- User agent: The pay button disappeared on my phone after choosing shipping — I had to scroll a lot.
- User agent: The OTP took 40 seconds to arrive — I almost left.
- User agent: Prices are clear and comparing plans is easy 👌
- User agent: Sign-up was clear and fast — done in under a minute 👍
- User agent: Search is fast, but the filters reset every time I go back.
- User agent: Demo shop example: the checkout omitted its currency label.
- 01Pick who tries itCountries, devices and segments — the users that matter to you.
- 02They use it like real peopleEach agent tries to get something done and narrates the friction.
- 03See a satisfaction scoreSentiment split, top friction and honest quotes.
- 04Vote on what to fix firstYour team ranks what users struggled with.
- 05Feed it into the next inspectionPreetest’s crew checks the spots users stumbled on.
Included in the Scale plan.
Proof
Show your quality.
Pass an inspection and your system earns a quality seal — with a QR code unique to it that opens its public report.
The site works: a functional pass, score 60+, nothing critical.
Interface, performance and accessibility at standard depth, score 80+, nothing critical.
All of the above plus security on a verified domain, score 90+, nothing critical or high.
Seal template — yours carries your score, your domain and your own QR code.
Sample
A QR code for every system
It opens that system’s signed public report — print it on your shop window and invoices, or put it in your site’s footer.
PreetestAdvanced 94
And a compact badge for README files.Pricing
Pay only for the work done.
One unit, the token: an estimate is reserved, then settled to real usage.
Free
$0
Try the platform
- Free tokens when you sign up
- 2 seats · 3 systems
- +30 tokens when you sign up — 50 in total
- All 17 agent specialties on Claude Sonnet
- Full report, fix prompts, retests
Starter
$29/mo
≈ 1 standard inspection a month
- 1,000 tokens a month
- Code scan alongside the screens
- 2 seats · 5 systems
Pro
$79/mo
≈ 4 standard inspections a month
- Everything in Starter, plus:
- 3,000 tokens a month
- 5 seats · 20 systems
- Pick the AI model · up to 2 per run
- GitHub, webhooks, Slack, API keys
Scale
$159/mo
≈ 10 standard inspections a month
- Everything in Pro, plus:
- 7,000 tokens a month
- 25 seats · 100 systems
- Up to 4 models per inspection
- Community and simulated-user fleet
Inspection counts are approximate: real token use depends on each site and is not 100% predictable — you are charged only what is actually used.
Prices in USD, charged in Saudi riyals (SAR 3.75 / $). Extra tokens can be bought inside any plan. + 15% VAT for customers in Saudi Arabia.
Enterprise: private deployment and SLA — sales@preetest.com
FAQ
Straight answers
When can I start?+
Now. Enter your site's link at the top of this page: and start your first inspection free, no card. Create a free account afterwards to keep the report and get more free tokens.
Can I cancel or change my plan?+
Yes, any time from Plans & tokens. Moving up takes effect at once and the unused days of the current plan are credited; cancelling stops the renewal and you keep access until the end of the period. A reminder arrives 3 days before each renewal.
Do unused tokens expire?+
No. Your plan's monthly tokens and any top-ups stay yours and carry over month to month. Each job reserves its estimate and charges only what it really used.
How do I pay?+
By card — mada, Visa or Mastercard — or STC Pay, through Moyasar, a payment gateway licensed by the Saudi Central Bank. Prices are in USD and charged in Saudi riyals (with VAT for customers in Saudi Arabia); an invoice reaches your e-mail.
Do I need a developer?+
No. Enter your address and get a report scored out of 100, with a fix prompt for every issue.
Is my site safe during an inspection?+
Yes. Agents are isolated from private networks, credentials are encrypted, secrets are redacted before storage, and signed-in scans require a verified domain.
Which AI model does the crew use?+
Claude Sonnet on every plan (the instant test runs on Claude Haiku). From Pro: other models enabled on the platform (GPT, Gemini, Grok, Llama, DeepSeek, Qwen, Mistral as available) — up to 2 models per inspection (4 on Scale).
What is a token?+
A unit measuring AI model usage plus browser time. Usage varies with the pages, crew, inspection depth and model; you pay only actual usage.
Your next release, without surprises.
Inspect your site now, free — no card.
✓Your place is reserved
#1
We sent you a welcome e-mail. Every friend who joins with your link moves you up 5 places.