AQAI - Quality Automation Artificial Intelligence

AQAI - Quality Automation Artificial Intelligence is a self-reported AI-assisted QA platform that claims to transform software requirements into verified, executable automated tests through an evidence-first pipeline. The author states that AQAI uses GPT-5.6 for semantic reasoning and deterministic Java generation for automation, with a focus on separating probabilistic AI from deterministic execution. It integrates UI discovery, evidence validation, knowledge reuse (via Neo4j and Qdrant), and structured test generation.

The project is presented as a vertical slice built during a hackathon, demonstrating end-to-end functionality including requirement processing, UI discovery, code generation, compilation, and test execution. The author emphasizes that AI reasoning is constrained by verified evidence and deterministic quality gates, aiming to avoid unrestricted LLM control over executable automation.

The description states that AQAI is a platform for AI-assisted QA that transforms software requirements into verified, executable automated tests using an evidence-first pipeline. It includes:

• A structured workflow: Requirements → Canonical Test Cases → UI Discovery → Verified Evidence → Confirmed UI Catalog → GPT-5.6 Reasoning → Typed Contracts → Deterministic Java Generation → Quality Gates → Test Execution.
• Use of GPT-5.6 for semantic tasks such as page enrichment and Page Object contract planning.
• Deterministic components responsible for UI evidence validation, locator scoring, Java Page Object generation, TestNG test generation, compilation, code review, smoke validation, and execution.
• Integration of Neo4j and Qdrant to support persistent knowledge reuse.
• Built with Java 17, Selenium WebDriver, TestNG, Maven, Docker, and other tools.

The author describes AQAI as a system where AI reasoning is constrained by curated evidence and validated contracts before influencing automation code. It is not a general-purpose AI tool but a specific QA platform built around an architecture that separates AI from deterministic execution.

Claimed Functionality

The product claims to generate executable test automation from requirements using AI-assisted semantic reasoning, while ensuring all generated code is grounded in verified UI evidence and deterministic engineering practices.

The description indicates a clear positioning shift from generic AI-driven test generation to a more structured, evidence-grounded approach. The author notes:

• Initial experiments with LLMs led to problems such as unverified locators, incorrect assertions, inconsistent patterns, and token inefficiency.
• This prompted the design of an evidence-first pipeline where verified UI evidence controls what becomes executable code.
• The system is positioned as a platform that combines semantic reasoning capabilities of modern language models with deterministic software engineering practices.

The evolution shows:

1. Initial Idea: AI should reason about software, but verified evidence should control automation.
2. Core Concept: AI-assisted QA that separates probabilistic reasoning from deterministic execution.
3. Key Innovation: Use of persistent knowledge layers (Neo4j/Qdrant) to avoid repeated semantic reasoning while maintaining quality gates.

Positioning Claim

AQAI positions itself as an AI-powered QA platform grounded in verified evidence, aiming to improve reliability and reproducibility in test automation.

The description does not explicitly state target customers or ideal customer profiles (ICP). However, based on the technical architecture and use case:

• The system is built for software engineering teams working with UI-based applications.
• It targets users who need automated test generation, particularly in environments where reliability and reproducibility are critical.
• The platform appears aimed at QA engineers or DevOps teams looking to integrate AI into their testing workflows.

Target Customer

Likely QA engineers, DevOps teams, or software development teams working with UI-based applications requiring reliable automated testing.

ICP Not Evidenced

No explicit ICP defined beyond the general audience of developers and QA practitioners using UI automation tools.

There is no evidence in the description regarding business model or pricing. The author only describes the technical architecture, pipeline stages, and demo outcomes.

Business Model

Not evidenced.

Pricing Evidence

Not evidenced.

The author provides detailed information about how AQAI was built:

• Technology Stack: Java 17, Selenium WebDriver, TestNG, Maven, GPT-5.6, Neo4j, Qdrant, Docker.
• Architecture: Staged pipeline with typed artifacts exchanged between stages; AI reasoning is constrained and validated before influencing deterministic code generation.
• AI Role: GPT-5.6 used for semantic tasks like page enrichment and contract planning.
• Deterministic Components: Responsible for UI evidence validation, locator scoring, Page Object generation, compilation, code review, smoke validation, and execution.
• Knowledge Reuse: Neo4j and Qdrant store validated application knowledge to reduce redundant AI calls.
• Demo Results: Successfully processed 4 requirements, generated 4 canonical scenarios, confirmed 2 pages, produced 2 Java Page Objects, 4 TestNG tests, passed compilation and smoke validation, and executed all tests successfully.

Technical Signals

Strong architectural separation between AI reasoning and deterministic execution; use of persistent knowledge layers for efficiency.

Delivery Signals

End-to-end demo completed with successful test generation and execution; demonstrated ability to reuse validated knowledge.

The description states that AQAI was built during a hackathon (OpenAI 2026) and includes a demo showing:

• Successful processing of 4 capability-first requirements.
• Generation of canonical test cases, UI discovery, evidence verification, Java Page Objects, TestNG tests.
• Compilation, code review, smoke validation, and execution of all generated tests.

However, there is no evidence of:

• Real-world usage or adoption beyond the demo.
• Revenue, customers, or traction data.
• Product-market fit or market validation.
• Scaling beyond a single-person development effort.

Traction Signals

Limited to a hackathon demo with no indication of real-world usage or commercial traction.

Maturity Signals

Demonstrated vertical slice functionality; architecture shows maturity in separating AI from deterministic execution, but lacks evidence of production readiness or scalability.

The description does not mention competitors or competitive positioning. It focuses solely on AQAI's internal design and functionality.

Competitive Context

Not evidenced.

Market Positioning

Not evidenced.

Several risks and open questions are implied by the self-reported nature of the project:

1. No Independent Verification: All claims are self-reported and unverified.
2. Single Developer Team: Only one member listed (PKrasytskyi Krasytskyi), suggesting limited development capacity or lack of team structure.
3. Demo-Only Scope: No evidence of real-world usage, customer feedback, or product-market fit beyond a hackathon demo.
4. Unproven Scalability: The architecture may work in a controlled demo but is untested at scale.
5. AI Dependency Risk: Heavy reliance on GPT-5.6 and AI reasoning without clear fallbacks or error handling for model failures.
6. Knowledge Layer Complexity: Integration of Neo4j and Qdrant introduces complexity that may not be scalable or maintainable in production.

Key Risks

Lack of independent verification, single developer team, demo-only scope, unproven scalability, dependency on AI models, and complexity of knowledge reuse layer.

Open Questions

No revenue, customers, or traction data; no indication of product-market fit or commercial viability.

1. What is the current status of AQAI beyond the hackathon demo? Has it been used in any real-world environments?
2. How does AQAI handle failures or edge cases during UI discovery or test generation?
3. Are there plans to support more complex application types (e.g., mobile apps, APIs)?
4. What are the limitations of the current knowledge reuse system? Can it be extended to support larger applications?
5. Is there any plan for monetization or commercialization beyond the demo?
6. How does AQAI ensure consistency and accuracy when reusing previously validated knowledge?
7. What is the expected performance impact of integrating AI reasoning with deterministic execution?
8. Are there any known issues with model hallucinations or incorrect outputs in the current implementation?