✨ v3.0 — AI-Powered Test Automation & AI Enablement Platform

Test Automation That
Thinks For Itself

The only testing tool that makes websites AI-compatible

Record tests by browsing. Generate them from plain English. Run across 5 browser options in parallel. When they break, AI heals them automatically. Export an MCP server and any AI can operate your website. 55+ features from visual regression to AI bridge export — the most complete AI testing platform available.

Start Free Trial See Screenshots
55+
Built-In Features
5
Browser Options
8x
Parallel Workers
3
AI Providers
1-Click
AI Bridge Export

See It In Action

Watch how AI transforms your testing workflow

🤖 NEW in v3.0 — Headline Feature

Turn Any Website Into an AI-Powered Tool

Record your workflows. Certify them. Click Export. Now any AI assistant — Claude, GPT, or others — can operate that website through natural language. No API required. No developer effort on the website side.

1
🎥

Record & Certify

Record user workflows on any website. Certify (lock) the ones that are production-ready. The AI auto-annotates each step with human-readable descriptions.

2
📦

Export MCP Server

Click the purple robot button. A standalone MCP server package downloads as a zip — complete with Playwright automation, tool definitions, and setup instructions.

3
🧰

Connect Any AI

Configure Claude Desktop (or any MCP-compatible AI client) with one JSON snippet. The AI can now run workflows, fill forms, navigate pages, and compose new actions.

Four Capability Layers

🔧

Workflow Tools

Each certified test becomes a callable tool with parameters. login(email, password), create_template(name, type), submit_form(name, email, message).

⚙

Atomic Tools

Low-level primitives (click, fill, navigate, screenshot, etc.) the AI composes for NEW actions not covered by recordings.

🔍

Task Analysis

analyze_task(description) fuzzy-matches a plain English request to the right workflow and returns mandatory field questions. No AI API key needed.

🌐

Site Knowledge

Scan data exposed as resources (site://pages, site://elements/{page}, site://mandatory-fields) so the AI knows every page, field, and validation rule.

✓ Works on ANY website — React, Rails, Angular, legacy apps, even sites with no API
✓ Zero effort from the website's developers
✓ Self-healing keeps the bridge working when the site changes
✓ Generated server is completely standalone — no dependency on AA Regression Tester at runtime
✓ No AI API key required in the exported package — task analysis is all local
✓ Mandatory field awareness — the AI knows which fields are required and asks before running

Watch: Making Apps AI Ready

See How AI Operates Your Website

User: “Create a new form template”
Claude: calls analyze_task → learns “Template Name” and “Template Type” are mandatory → asks the user
Claude: “What should the template be called, and what type? Options: Quarterly, Monthly, Annual”
User: “Call it Q1 Report, make it Quarterly”
Claude: runs create_template(name=“Q1 Report”, type=“Quarterly”)
Result: ✔ Template created successfully
User: “Create 10 templates called Template A through Template J”
Claude: reads site://mandatory-fields to learn required fields, then loops through navigate → fill → click sequences 10 times
Result: ✔ 10 templates created successfully

See the Platform in Action

Real screenshots from AA Regression Tester v3.0

AA Regression Tester Dashboard with 3-column layout showing test library, browser preview, and visual diff viewer

3-Column Glassmorphism Dashboard

Manage your test library with folders, tags, and search on the left. Watch live browser previews and console output in the center. Review visual diffs, run history, and analytics on the right.

Live Preview Visual Diffs Run History Analytics Chart
AI Test Generator with natural language input, AI Do It button, and Auto Smoke Tests panel

AI-Powered Test Generation

Describe what you want to test in plain English. Choose from "AI Do It" for instant generate-and-run, or use Auto Smoke Tests to create navigation, form, dropdown, and button tests automatically from your scanned elements.

Natural Language AI Do It Auto Smoke Tests Field Scanner
Suite Management with test suites, schedules, pass/fail results, and parallel execution

Suite Management & Scheduling

Organize tests into suites and run them sequentially or in parallel with up to 8 workers. Schedule runs with cron expressions. View live progress, pass/fail badges, duration tracking, and suite history at a glance.

Parallel Execution Cron Schedules Live Progress Pass/Fail Badges
Settings modal showing AI provider selection, API key, visual threshold, browser engine, and display mode

Flexible Configuration

Choose your AI provider (Gemini, OpenAI, or Claude), configure visual regression thresholds, select browser engine and display mode, set up parallel workers, and manage notification channels — all from a single settings modal.

3 AI Providers 5 Browser Engines Visual Thresholds Notifications

Core Capabilities

Six pillars of autonomous test automation

AI Self-Healing & Root Cause Analysis

When selectors break, AI analyzes the page DOM and proposes corrected selectors. When healing fails, structured Root Cause Analysis explains why — with categories like selector_changed, timing_issue, or app_error.

  • Multi-provider: Gemini, OpenAI, Claude
  • Auto-retry with flaky detection badges
  • Smart failure classification before AI spend

Visual & AI Assertions

SSIM-based pixel comparison catches UI regressions. Visual AI assertions let you describe what the page should look like in plain English. Natural language assertions validate logic without writing selectors.

  • SSIM visual regression with configurable thresholds
  • AI visual checks via screenshot analysis
  • Natural language assertions for any validation

Record, Edit & Replay

One-click recording from any URL with smart element detection. Save checkpoints mid-session. Edit steps with drag-and-drop reordering. Replay simulates human-like mouse movements and typing with realistic pacing.

  • Recording checkpoints & password redaction
  • Step editor with inline editing & reordering
  • TinyMCE rich text editor support

AI Test Generation & Smoke Tests

Describe what you want to test in natural language. AI generates complete test cases using real selectors from your scanned elements. "AI Do It" generates and immediately runs the test in one click.

  • Natural language to executable test cases
  • 4 auto smoke test templates
  • Field Scanner for real selector extraction

Data-Driven & Conditional Testing

Run tests N times with different inputs from CSV or JSON data files. Add conditional branching with if_exists/if_not_exists logic, wait-for conditions, and network assertions to validate API calls.

  • CSV/JSON parameterization with {{variable}} syntax
  • Conditional branching with nested logic
  • Network assertions & API mocking

Parallel Execution & CI/CD

Run suites across Chromium, Firefox, WebKit, Chrome, and Edge with up to 8 parallel workers. Integrate with any CI/CD pipeline via webhooks, JUnit XML export, and scheduled cron runs.

  • 1–8 concurrent workers across 5 browsers
  • Slack, Teams & email notifications
  • Bearer token auth & webhook callbacks

55+ Features. One Platform.

Everything you need for autonomous test automation & AI enablement

🎥

Record from URL

Enter a start URL and the browser navigates there automatically before recording begins. Waits for page load and network idle.

💾

Recording Checkpoints

Save actions recorded so far as a named test case without stopping the recording. Create incremental test snapshots on the fly.

✎

Step Editor & Reordering

Inline-edit selectors and values. Drag-and-drop steps to reorder. Add waits, conditions, and network assertions with dedicated buttons.

🔒

Password Redaction

Password fields are automatically detected during recording. Actual values are replaced with {{PASSWORD}} placeholders to prevent credential leaks.

📝

TinyMCE Support

Specialized recording and replay for TinyMCE rich text editors. Handles iframe-based content, dynamic IDs, and multi-strategy interaction.

⌨

Keyboard Shortcuts

Ctrl+Shift+R to record, Ctrl+Shift+S to stop, Ctrl+Shift+P to play, Ctrl+Shift+Space to pause/resume. No mouse required.

🎦

Video & Step Screenshots

Record WebM video of every replay. Capture PNG screenshots after each step for a filmstrip view. Debug exactly what the browser saw.

🔎

Element Highlighting / Locate

Click “Locate” on any step to highlight the element in the live browser. Visual confirmation that selectors are pointing at the right target.

➕

Manual Step Addition

Add steps manually without recording — click, fill, assert, navigate, wait, and more. Full step type picker with selector and value fields.

🔄

Change Button (Re-pointing)

3-phase in-browser UX: navigate, hover-to-select, then confirm. The test step updates itself with the new selector and replay continues.

📂

Step Categories

Steps are tagged as NAV, ACTION, ASSERT, or CTRL with color-coded badges. Collapsible test settings panel for URL, tags, and metadata.

🤖

Multi-AI Self-Healing

When selectors break, AI analyzes page context and proposes corrected selectors. Supports Gemini, OpenAI, and Claude with configurable retry attempts.

💬

AI Test Generation

Describe tests in natural language. AI generates executable test cases with real selectors from your scan data and existing test library as context.

⚡

AI Do It

One-click generate-and-run. Describe what to do, and the tool generates the test, saves it, and runs it in the browser immediately.

🚬

Auto Smoke Tests

Generate smoke tests from scan data with 4 built-in templates: navigation links, fill all forms, open all dropdowns, and click all buttons.

🔍

Field Scanner

Automatically catalog all interactive elements as you browse. Detects buttons, links, inputs, selects, MUI components, and more.

🔬

AI Root Cause Analysis

When healing fails, AI provides structured analysis: root cause, suggestion, and category (selector_changed, timing_issue, app_error, etc.).

👁

Visual Regression (SSIM)

Screenshot comparison using Structural Similarity Index. Configurable thresholds (0.90–1.00). Side-by-side diff viewer with baseline, current, and diff images.

💎

Visual AI Assertions

Describe what the page should look like in natural language. AI analyzes a screenshot and returns pass/fail with an explanation. Supports all 3 AI providers.

🧠

Natural Language Assertions

Write any assertion in plain English: "The total price should be greater than $0". AI evaluates both screenshots and DOM context for judgment.

🌐

Network Assertions

Validate that specific HTTP requests occurred during replay. Match by URL pattern, HTTP method, and expected status code.

🔗

Network Mocking

Define mock API responses with custom status codes, bodies, headers, and artificial delays. Isolate frontend tests from backend dependencies.

⏳

Wait-For Conditions

Wait for elements (visible, hidden, attached, detached), URL patterns, or text appearance with configurable timeouts. Essential for dynamic UIs.

🔀

Conditional Branching

Add if_exists / if_not_exists conditions with then/else action arrays. Supports nested conditions for complex test flows.

📊

Data-Driven Tests

Attach CSV or JSON data files. Tests run N times with different inputs using {{variable_name}} syntax in selectors and values.

🌍

Multi-Environment

Configure dev, staging, and production environments. URLs are automatically swapped during replay. CLI supports --env flag for headless runs.

🎯

Interactive Re-pointing

3-phase in-browser UX: navigate to find the element, hover to select it, then confirm. The test updates itself and replay continues.

⏱

Performance Baselines

Per-step timing recorded during every replay. Timing badges flag slow steps (>2s). Historical trend data via API for the last 20 runs.

📁

Folders, Tags & Search

Organize tests into subfolders with drag-and-drop. Add tags like "smoke" or "critical". Instant search filters by path and name with 300ms debounce.

☑

Bulk Operations

Enter bulk mode to select multiple tests. Run, move, tag, or delete them all at once. Exit bulk mode to return to normal view.

📜

Versioning & Changelog

Every modification is recorded: manual edits, AI healing, re-pointing, bulk changes. View full change history with timestamps and icons.

📈

Analytics Chart

Doughnut chart showing aggregated passed, failed, and healed test counts. Overall pass rate percentage updates automatically after each run.

📄

JUnit XML & HTML Reports

Export results as JUnit XML for CI/CD pipelines or self-contained dark-theme HTML reports. Available via API, CLI, or dashboard download.

📅

Suite Management & Scheduling

Create suites, run them with live progress tracking, schedule via cron or intervals. Suite cards show pass rate, duration, and mini colored bar.

🌐

Multi-Browser Support

Chromium, Firefox, WebKit, Google Chrome, and Microsoft Edge. External (visible) or embedded (headless) display modes.

⚡

Parallel Execution

1–8 concurrent workers via ThreadPoolExecutor. Isolated browser contexts for each test. Results preserved in original order.

📦

Docker Support

Dockerfile and docker-compose.yml included. Python 3.11-slim with Playwright pre-installed. Volumes for persistent test data and configuration.

🔒

Authentication (PBKDF2)

Built-in login with PBKDF2-SHA256 hashing (260,000 iterations). Secure session management with HttpOnly, SameSite, and Secure cookies.

🔐

Secret Encryption (Fernet)

API keys and sensitive values encrypted at rest using Fernet symmetric encryption. Decryption happens transparently on load.

🛡

Security Headers

CSP, X-Frame-Options: DENY, X-Content-Type-Options: nosniff, strict Referrer-Policy, and Permissions-Policy on all responses.

🔔

Notifications

Slack webhooks, Microsoft Teams MessageCards, and SMTP email. Configurable: notify always, on failure only, or never. Channels are independent.

🔃

Import & Export

Import Selenium IDE .side files. Export standalone async Playwright Python scripts. Full round-trip compatibility with existing test ecosystems.

🎯

Recording-Time Selector Validation

Every selector tested against the live DOM during recording. Auto-upgrades to the most unique selector available. Purple toast notifications show improvements in real-time.

🔎

MUI & Portal Dropdown Detection

Automatically detects MUI, Bootstrap, Ant Design, Vuetify, Radix, Chakra, and 10+ other framework dropdowns — records them as proper selections, not brittle clicks.

🏷

Element Label Capture

Captures human-readable labels for every element so tests are self-documenting. Labels are used in AI annotation and MCP export descriptions.

🔢

Duplicate ID Handling

When multiple elements share the same ID (common in Bootstrap), the recorder detects it and uses alternative selectors automatically.

📏

Precise Drop Positioning

Drag-and-drop captures exact pixel offsets, not just center-of-target. Ensures pixel-perfect replay of complex drag interactions.

📐

Resize Handle Detection

Detects and records resize handle interactions with precise positioning for panel and element resize operations.

🔒

Certified / Locked Tests

Lock tests to prevent accidental modification. AI auto-annotates each step on certification. Bulk operations automatically skip certified tests.

🤖

AI Annotation on Certification

When you certify a test, AI adds human-readable descriptions, interaction types, and field labels to every step — making exports and reports self-documenting.

📈

Full Analytics Dashboard

4 KPI cards (Pass Rate, Total Runs, Avg Duration, AI Healed), trend charts, top failures, slowest tests, and AI healing success rate.

🛑

Stop Replay Mid-Run

Stop any test at any time — responsive even during AI healing or re-pointing. No more waiting for a broken test to time out.

🔔

Alert / Overlay Capture

DOM alerts, toasts, and notifications are automatically captured during replay with type classification (danger, warning, success, info).

🔄

Smart Retry (AI Do It)

AI-generated tests that fail with validation errors are auto-corrected using failure screenshots and error messages — up to 3 attempts.

👁

Visual Overlap Check

AI vision checks after drag-drop sequences to detect overlapping elements. Catches UI issues that pixel comparison misses.

🌐

MCP Export (Website AI Bridge)

Export certified tests as a standalone MCP server with analyze_task tool and site://mandatory-fields resource. Any AI can operate your website.

🤖

Internal MCP Server (17 Tools)

17 AI-callable tools for controlling the regression tester via Claude or any MCP client. Run tests, view analytics, generate tests, manage suites.

📈

Analytics with Trend Charts

Full analytics tab with date range filters, trend charts, top failures, slowest tests, and AI healing success rate.

⭐

Mandatory Steps

Mark any test step as mandatory with a gold star toggle. AI generation always includes them verbatim. Bundled into MCP Export for structured AI questions.

💬

Conversational AI Do It

AI asks clarifying questions about mandatory fields, dropdown values, and test data before generating tests. Question types include free text, option chips, and multi-select.

💫

Graceful Degradation

If AI analysis fails, it falls through to direct execution — no crash, no delay. “Skip Questions” button for users who want immediate execution.

How It Works

Five steps to autonomous testing & AI enablement

01

Record or Generate

Browse your application while recording — every click, type, drag, and navigation is captured with smart element detection. Or describe what to test in plain English and let AI generate the test case. Save checkpoints mid-recording. Passwords are auto-redacted.

Smart Selectors AI Generation Checkpoints Password Redaction
02

Organize & Configure

Organize tests into folders with tags and search. Set up suites with cron schedules. Configure multi-environment URLs so the same tests run against dev, staging, and production. Add data-driven parameters, conditional logic, and network mocks.

Suites & Tags Data-Driven Conditions Multi-Environment
03

Run in Parallel

Execute tests across Chromium, Firefox, WebKit, Chrome, and Edge with up to 8 parallel workers. Get video recordings, per-step screenshots, and performance baselines for every run. Watch live progress in the dashboard or run headless via CLI.

5 Browsers 8x Parallel Video Recording Performance Timing
04

AI Heals, Reports & Bridges

When selectors break, AI automatically repairs them. Flaky tests are detected and flagged. Get notifications via Slack, Teams, or email. Export JUnit XML for CI/CD — or export an MCP server that turns your tested website into an AI-operable application.

Self-Healing Flaky Detection Notifications JUnit XML MCP Export
05

Make It AI-Compatible

Certify your best tests, click Export MCP Server, and hand the zip to any team that wants AI assistants to operate the website. The generated server includes workflow tools, atomic actions, and full site knowledge — ready for Claude Desktop in minutes.

Certified Tests MCP Server Workflow Tools Site Knowledge

Enterprise-Grade Security Built In

From encrypted secrets to hardened headers, security is the foundation — not an add-on

AA Regression Tester runs entirely on your local machine — no cloud service required. Authentication, encryption, and security headers are built into every layer, from login to API access to test data storage.

AA Regression Tester secure sign-in page with dark glassmorphism theme
✓
PBKDF2 Password Hashing

SHA-256 with 260,000 iterations and random 32-byte salt per user

✓
Fernet Encryption at Rest

API keys and secrets encrypted with symmetric Fernet encryption in config.json

✓
Password Redaction

Password fields auto-detected during recording and replaced with {{PASSWORD}} placeholders

✓
Security Headers

CSP, X-Frame-Options: SAMEORIGIN, X-Content-Type-Options, strict Referrer-Policy

✓
Secure Sessions

HttpOnly, SameSite=Lax, and Secure cookie flags with randomly generated secret keys

✓
API Token Authentication

Bearer token auth for CI/CD endpoints with timing-safe comparison

✓
Docker Deployment

Containerized with docker-compose. Persistent volumes for test data and config

✓
Multi-Environment Isolation

Separate dev, staging, and production configurations with automatic URL swapping

✓
MCP Bearer Token Auth

Bearer token authentication on all API routes — MCP servers and external tools authenticate seamlessly

✓
Certified Test Protection

Certified/locked tests prevent accidental modification of production-verified workflows

How We Compare

The most comprehensive testing platform on the market

Capability Selenium / Cypress Cloud Platforms AA Regression Tester
Self-healing✘ NoBasic (DOM)✔ AI-powered (3 providers)
Natural language assertions✘ No✘ No✔ Yes
Visual AI✘ NoExpensive tier✔ Built-in
Root cause analysis✘ No✘ No✔ AI-generated reports
MCP / AI Bridge✘ No✘ No✔ 1-click export
Conversational AI generation✘ No✘ No✔ AI asks before generating
Mandatory field enforcement✘ No✘ No✔ Mark & preserve
Analytics dashboard✘ NoBasic✔ Full trend analysis
Certified tests✘ No✘ No✔ Lock + AI annotation
Smart retry✘ NoBasic✔ Screenshot-based correction
Self-hosted✔ Yes✘ No✔ Full data control
Framework detection✘ No✘ No✔ MUI, Bootstrap, Ant, 10+
CostFree (DIY)$300-2000/mo✔ Competitive

Built on Modern Technology

Enterprise-grade stack for reliability and performance

🤖
Multi-AI Engine
Gemini, OpenAI, Claude
🎭
Playwright
Chromium, Firefox, WebKit, Chrome, Edge
🐍
Python 3
Async/Await Architecture
⚡
Flask
RESTful API Server
👁
OpenCV + SSIM
Visual Regression Detection
🔐
Fernet Encryption
Secrets Encrypted at Rest
📦
Docker
Container Deployment
📅
APScheduler
Automated Test Scheduling
🔗
MCP Protocol
FastMCP for AI Integration
📈
Chart.js
Analytics Visualizations

Works With Your Stack

Browsers, CI/CD, notifications, and import/export compatibility

Browsers

Chromium Firefox WebKit Google Chrome Microsoft Edge

CI/CD & Automation

Webhook API REST Endpoints Scheduled Cron Runs Docker Compose CLI with JSON Output MCP Server (Internal — 17 tools) MCP Export (Standalone AI Bridge)

Import & Export

Selenium IDE Import Playwright Script Export JUnit XML Reports CSV/JSON Data Files MCP Server Export (.zip)

Notifications

Slack Webhooks Microsoft Teams SMTP Email

Frequently Asked Questions

Everything you need to know before getting started

Does my website need to have an API?

No. The tool operates through the browser UI, just like a human would. Any website that works in a browser works with AA Regression Tester.

Which AI provider should I choose?

All three (Gemini, OpenAI, Claude) work well. We recommend Claude for the best natural language understanding, but you can switch providers anytime.

Can I use this with my existing Selenium tests?

Yes! Import .side files from Selenium IDE and they're converted automatically. You can also export Playwright Python scripts for use in your existing CI/CD pipelines.

What does the MCP Export do exactly?

It packages your certified test workflows into a standalone server that any AI assistant can use to operate your website — no changes to the website needed. It even includes an analyze_task tool that guides the AI through mandatory fields.

Does the exported MCP server need an AI API key?

No. The exported server uses Playwright for browser automation. The analyze_task tool does local fuzzy-matching — no AI API calls needed. The AI client (Claude, GPT) provides its own intelligence.

What are mandatory steps?

You mark certain test steps (like required form fields) as mandatory. The AI generation system always includes them, and the exported MCP server knows to ask about them before running workflows.

Is my data sent to the cloud?

Only AI API calls (using your own key) leave your machine. All test data, recordings, and configuration stay local. Full self-hosted data control.

How is this different from Testim or Mabl?

We're self-hosted (full data control), support 3 AI providers, have natural language assertions, AI root cause analysis, conversational AI generation, mandatory step enforcement, and the MCP Export — none of which competitors offer.

Simple, Transparent Pricing

Choose the plan that fits your team

Starter

$ 49 /month

Perfect for small teams and side projects

  • ✓ Up to 50 test cases
  • ✓ 100 test runs/month
  • ✓ AI self-healing (single provider)
  • ✓ Visual regression detection
  • ✓ Record & replay
  • ✓ Single browser engine
  • ✓ Email notifications
  • ✓ Web dashboard
Get Started
Most Popular

Professional

$ 149 /month

For growing teams with serious testing needs

  • ✓ Unlimited test cases
  • ✓ 1,000 test runs/month
  • ✓ All 3 AI providers (Gemini, OpenAI, Claude)
  • ✓ Visual + AI assertions
  • ✓ AI test generation & Auto Smoke
  • ✓ 5 browsers + 4 parallel workers
  • ✓ Data-driven & conditional testing
  • ✓ Slack, Teams & email notifications
  • ✓ CI/CD integration + JUnit XML
  • ✓ Video recording & step screenshots
  • ✓ MCP Export (Website AI Bridge)
  • ✓ Analytics dashboard
  • ✓ Priority support
Get Started

Enterprise

Custom

For organizations requiring custom solutions

  • ✓ Everything in Professional
  • ✓ Unlimited test runs
  • ✓ 8 parallel workers
  • ✓ Docker / on-premise deployment
  • ✓ Fernet secret encryption
  • ✓ SSO & SAML
  • ✓ Multi-environment management
  • ✓ Internal MCP Server (17 tools)
  • ✓ Custom AI model configuration
  • ✓ Dedicated support & SLA
  • ✓ Training & onboarding
Contact Sales

Ready to Let AI Handle Your Testing?

Join teams using the most complete AI-powered test automation platform

⚡ Get Started