Automated Testing AI

Vision AI drafts your test steps, then runs them with Puppeteer

Intermediate
Screenshot of Automated Testing AI, Vision AI drafts your test steps, then runs them with Puppeteer

What is Automated Testing AI?

Automated Testing AI is a developer tool. Atai uses vision AI to draft automated test steps from a written description, then reruns those saved steps with Puppeteer at no token cost. A $99 lifetime download with full source code or a $19 per month cloud subscription, currently Mac only in beta.

Automated Testing AI (Atai, pronounced ah-tay) is a test automation workspace built around a digital assistant called Sprucebot. Rather than writing selector logic and helper functions by hand, you describe a test case in plain language: what the flow should do, what counts as success, and which edge cases matter. Sprucebot passes that prompt to a vision-capable model, watches your app render, and drafts the individual test steps on your behalf. Each suite is anchored to an endpoint, so the same workspace can hold separate collections for different apps or environments. The design choice that separates Atai from most AI test generators is where the model stops. Vision AI is used once, at authoring time, to produce steps that are then saved and executed directly with Puppeteer. Repeated runs do not consume tokens, so regression cycles stay cheap, and you can edit the generated steps by hand instead of returning to the prompt. Dummy variables and dummy users let you pin reusable credentials, addresses and other fixtures so tests stay predictable across runs. Screenshots of every step are pushed to your own S3 bucket for later review, and a ride along mode shows Sprucebot building a test in real time so you can interrupt a run that drifts and adjust the prompt. When a UI change breaks a test, Sprucebot can be asked to repair it against the new design. Atai suits QA engineers and small product teams who already understand their application's critical paths but resent the upkeep of a test suite. Cost passthrough is central to the pitch: you supply your own OpenAI key and S3 bucket, so platform pricing stays separate from model spend. Two routes exist, a $99 one time download that includes full source code and developer documentation, and a $19 per month cloud subscription. The project is in beta from a two engineer studio, Mac only for now, with Windows and Linux access available on request. Comparisons will inevitably land on hosted AI testing platforms. Atai trades their polish and breadth for transparency: the code is yours, the storage is yours, and the execution engine is a well known open source tool rather than a black box.

How do you use Automated Testing AI?

  1. 1Choose a lifetime download or a cloud account
    Pick the $99 one time download if you want to run Atai locally with source access, or create a cloud account at $19 per month. Both routes lead to the same test building interface.
  2. 2Connect your OpenAI key and storage
    Atai passes model and storage costs through to you, so add a vision capable OpenAI key and point screenshot uploads at your own S3 bucket before building anything.
  3. 3Create a test suite for an endpoint
    Suites group test cases and let you target a specific starting endpoint, which keeps separate apps or environments from mixing inside one workspace.
  4. 4Describe the test case to Sprucebot
    Write the flow you want covered, the success criteria, and any edge cases. Add dummy variables and dummy users so logins and addresses stay identical on every run.
  5. 5Build and watch the steps generate
    Press Build Test Steps and follow along in real time. If the assistant misreads the screen, stop it and refine the prompt before it finishes.
  6. 6Run, then repair when the UI shifts
    Saved steps replay without further model calls. When a redesign breaks a test, hand it back to Sprucebot for repair and review the step screenshots in your bucket.

Pros and cons

Pros

  • Test steps are generated once and replayed with Puppeteer, so repeat runs do not burn model creditsAI
  • Full source code and developer documentation come with the $99 lifetime downloadAI
  • Bring your own OpenAI key and S3 bucket keeps platform pricing low and spending transparentAI
  • Dummy users and variables make tests reproducible across runsAI
  • Ride along mode lets you watch generation live and stop a run that is going wrongAI

Cons

  • Only OpenAI vision models are supported at launch, so there is no model choice yetAI
  • Official support covers Mac with M series chips only; Windows and Linux access needs a manual requestAI
  • You must supply your own API key and storage bucket, which adds setup work and a second billAI
  • Downloaded copies cannot be hosted publicly, which rules out shared team servers without a commercial licenceAI
  • Beta stage from a two engineer team means fewer integrations and less polish than established suitesAI

How much does Automated Testing AI cost?

Pricing

Two plans. Download for Life costs $99 one time, runs locally on Mac, and includes the source code plus developer documentation. The Cloud Subscription is $19 per month. Model tokens and S3 storage are billed separately because you supply your own OpenAI key and bucket.

Learn more

Features

Vision AI drafts test cases from a written prompt, and the generated steps are saved and executed with Puppeteer so reruns cost nothing in tokens. Also included: unlimited suites and test cases, dummy variables and users, per step screenshots to your own S3 bucket, real time run monitoring, test repair after UI changes, ride along generation, and full source access on the lifetime plan.

Frequently asked questions about Automated Testing AI

  • How much does Automated Testing AI cost?
    Two plans. Download for Life costs $99 one time, runs locally on Mac, and includes the source code plus developer documentation. The Cloud Subscription is $19 per month. Model tokens and S3 storage are billed separately because you supply your own OpenAI key and bucket.
  • How do you use Automated Testing AI?
    The walkthrough on this page covers 6 steps: 1. Choose a lifetime download or a cloud account 2. Connect your OpenAI key and storage 3. Create a test suite for an endpoint 4. Describe the test case to Sprucebot 5. Build and watch the steps generate 6. Run, then repair when the UI shifts.
  • What platforms does Automated Testing AI support?
    Automated Testing AI is available on Web App and MacOS.
  • What are the limitations of Automated Testing AI?
    Only OpenAI vision models are supported at launch, so there is no model choice yet. Official support covers Mac with M series chips only; Windows and Linux access needs a manual request. You must supply your own API key and storage bucket, which adds setup work and a second bill.

Status

StatusActive
Views0
Outbound clicks0
Added10/21/2026

Platforms

Web AppMacOS

Pricing

PaidSubscription