RaptorTest turns a web application, its documentation, or both into reviewable test cases, automated test runs, and traceable evidence. Your QA team defines the strategy and stays in control.
Results from an actual test run · 2 passed, 2 failed, 1 blocked
Automation speeds up testing. QA strategy and professional judgment remain essential.
01
From application to test cases
Discovers web workflows and combines observed behavior with functional documentation to build test coverage that your team can review.
02
Human-supervised execution
Runs tests in a real browser, following the strategy and permissions defined by your QA team.
03
Evidence for informed decisions
Captures results, screenshots, and assertions so teams can assess findings and rerun test suites.
The evolution of software quality
From QA to QE. Now toward Quality Intelligence.
Software quality is no longer just about running tests. It also means understanding what to test, applying intelligence throughout the process, and turning results into information that supports better decisions.
RaptorTest brings together AI-assisted test design, automated execution, traceable evidence, and professional QA review.
QA
Quality Assurance
Set quality standards and verify that they are met.
QE
Quality Engineering
Build quality and automation into software development.
QI
Quality Intelligence
Connect context, testing, and evidence to make better decisions.
The challenge
More coverage, less manual test preparation
As a product grows, documenting test cases and keeping regression coverage current can become a bottleneck.
Applications with little or no test documentation.
Technology migrations that require functional test coverage to be rebuilt.
Teams spending too much time writing and maintaining test cases.
Releases that need fast validation of critical user journeys.
Complex workflows that require exploration, evidence, and follow-up.
Guides, PDFs, and screenshots that have not yet become reusable test suites.
Selenium or Playwright projects that still depend on prewritten test cases.
How it works
Seven stages, one supervised test run
These screenshots show an actual RaptorTest run on a sample application featuring project creation, a shopping cart, and two-factor registration.
STEP 1
Takes in the context
Your team provides the application, environment, testing mode, objectives, and business rules. You can also attach PDFs, images, and functional screenshots.
New test run. Mode, objective, documents, viewport, and test case count.
STEP 2
Explores the application
Navigates the site, inspects screens, controls, and forms, records routes, and groups business workflows without assuming functionality that the evidence does not support.
Live activity. Current stage, events, and latest browser screenshot.
STEP 3
Proposes a testing strategy
After exploring the application, it suggests approaches such as critical paths, validation and boundary tests, recovery scenarios, or broader coverage. The QA professional selects and combines options and adds instructions.
Strategy. “3 user journeys identified across 3 screens.”
STEP 4
Builds a reviewable test suite
Each test case includes preconditions, steps, expected results, priority, dependencies, source, and verifiable acceptance criteria. Cases can be edited, expanded, excluded, or saved without running them.
Test case editor. The QA professional can adjust any field before approval.
STEP 5
Executes tests in a real browser
Playwright performs actions step by step. Each test case starts from a controlled session and is checked against criteria defined before execution.
Approval. The suite does not run until QA approves it.
STEP 6
Asks for help when needed
If credentials, a verification code, authorization, or information known only to a person are needed, the agent pauses and asks rather than making anything up.
CredentialsOTPAuthorizationBusiness data
Human input required. The agent pauses and logs the request.
STEP 7
Produces evidence and reports
Records actions, screenshots, assertions, console errors, relevant HTTP responses, AI usage, and a chronological decision log.
Allure HTML. Self-contained technical report for each run.
A real finding
A defect found without writing a single test script
In the run shown above, the agent independently designed a shopping cart boundary test and discovered that the cart accepts negative quantities.
Failed
TC-004 · Enter a negative quantity and verify that the total does not change
Priority
Medium · Validation · Observed screens
Action
Enter -1 in Quantity and click Update
Expected
The total remains Total: 10
Observed
Total: -10
QA verdict
Confirmed as a genuine defect, with a note added to the findings report
Test case details. Planned steps, actual execution, assertions, and human verdict.
Three testing modes
Your team chooses the depth of testing
Smoke · 8 suggested cases
Smoke
Quickly checks that essential workflows operate as expected: page loading, main navigation, authorized sign-in, and core actions with visible outcomes.
Smoke test on saucedemo.com: sign-in and inventory checks passed.
Exploratory · 18 suggested cases
Exploratory
Expands the test scope to uncover risks: positive and negative paths, boundary cases, validations, empty states, recovery, persistence, and technical signals.
Cases grouped by user journey, including happy paths and validations.
Regression · reusable suite
Regression
Reruns an approved suite with the same steps and criteria. Compares previous and current results without generating new cases. In this example: 0 AI calls.
Regression rerun without consuming tokens.
Two ways to get value
Document first or execute right away
Design test cases without executing them
Ideal for legacy systems, establishing test coverage, or preparing a migration. The suite can be reviewed, edited, downloaded, and executed later. It can be built from:
explored application screens;
business rules;
one or more PDFs;
images or screenshots;
a combination of documentation and the actual application.
Suite built from a PDF. 6 cases organized into 3 user journeys, with no environment URL.
Design and execute
Ideal for releases, smoke testing, exploratory testing, and regression. The agent explores, plans, executes, verifies, and documents results within a single workflow.
Overview. 30-day activity, findings awaiting review, and AI spend.
Core capabilities
What the agent can do
Autonomous exploration of websites and web applications.
Exploration with either no fixed screen limit or a defined maximum.
Identification of user journeys and their dependencies.
Prioritization by impact and likelihood.
Happy paths, negative scenarios, boundary values, validations, alternative flows, and recovery.
Sign-in using an authorized test account.
Interactive requests for credentials, OTPs, or missing information.
Data creation and modification only when write access is authorized.
Full suite review and editing before execution.
Real-time visual evidence for individual steps.
Assertions for text, URLs, visibility, counts, values, titles, HTTP responses, and HTML5 validity.
Diagnostics for console errors, JavaScript, and relevant HTTP requests.
Automated accessibility checks with axe-core.
Application-level memory to prioritize recent failures and instability.
Reusable test suites to reduce planning overhead.
Tracking of tokens, latency, cost, and quality by model.
Interchangeable AI providers using the client’s own API key.
People remain in control
A supervised agent, not a black box
Define objectives and rules
Choose mode and budget
Select the strategy
Allow or prevent write operations
Review each test case
Edit steps and expected results
Add or remove assertions
Include, exclude, or remove cases
Respond to live requests
Stop a running test
Confirm or dismiss findings
Add notes to reports
A passed test means that its stated, executed criteria were met. It does not certify the entire application.
QA review. The human verdict supplements the technical result without overwriting it.
Evidence for QA and management
Export every run in the format your team needs
Evidence. Screenshot gallery organized by screen and test case.AI comparison. Actual quality, latency, and cost by provider and model.
PDFA structured, presentation-ready functional test suite.
MarkdownEasily editable test cases and findings.
CSVTabular processing and migration to other tools.
JSONComplete structured report.
JUnit XMLIntegration with CI pipelines and automation tools.
Allure HTMLTechnical report with test cases, steps, statuses, and attachments.
PrintPrint-ready view that can also be saved as a PDF.
Organizations, AI, and security
Isolated workspaces, flexible AI, and responsible use
Organization-based workspaces
Each client has an isolated workspace. Organization-level applications are shared with the team, while personal applications are visible only to their creators. Operators can run and review tests; Viewers can browse and export results. There is no public sign-up: BairesQA provisions every account.
Flexible AI and cost controls
Groq, OpenAI API, Anthropic Claude, Gemini, Kimi, or OpenRouter, using the client’s own API key. Set a budget per run and optional limits on time, calls, tokens, or steps. API keys are not stored in the browser, history, or reports.
Security
Authenticated sessions, hashed passwords, redaction of known secrets, per-run allowed domains, write operations blocked by default, and access to private networks blocked unless explicitly configured. No real payments or destructive actions during exploration.
AI engine. Provider, model, temporary API key, and budget.BairesQA administration panel. Users, roles, and permissions by organization.
Use cases
Where it delivers the most value
Migrating a legacy application
Turn existing guides and real user journeys into a test baseline before migrating, then run the same coverage against the new version.
Release validation
Run smoke tests for sign-in, navigation, and core functionality; quickly review failures, blockers, and evidence.
Assisted functional exploration
Explore modules and forms, looking for validations, boundary conditions, and recovery paths based on an agreed strategy.
E-commerce
Search, filters, product pages, carts, sign-in, and checkout up to the authorized limit, with no real payments.
Internal business applications
Record creation and editing, filters, permissions, persistence, and dependencies between entities, using identifiable QA test data.
Discover RaptorTest
A different approach to QA
Schedule a demonstration to see how RaptorTest designs, runs, and documents AI-assisted web tests while keeping your team in control.
RaptorTest does not offer automatic public sign-up. BairesQA sets up each client’s organization and user accounts. Schedule a demo to explore the platform and assess whether it fits your needs.
What RaptorTest does not replace
The client’s own business requirements.
Professional review of defects and false positives.
Specialized security, performance, or load testing.
A strategy for test data and environments.
Native mobile app automation (currently: desktop web and mobile web).
