AI QA agent
Point it at a product.
Get a map of it.
AI‑Lens explores a live application, builds a knowledge graph of its screens, journeys and business rules, then designs test cases from what it saw.
Watch it work
See AI‑Lens explore a product.
A recorded walkthrough: AI‑Lens explores a live application, builds its knowledge graph and designs test cases from what it found.
What AI‑Lens is
One QA agent, any product, any model.
AI‑Lens walks a live application, records what it finds in a single knowledge base, and designs, generates and runs tests from that record. It works with the tools your team already uses and the automation framework you already standardise on.
Reads the real product
Tests reference controls and flows it observed, not ones it assumed.
One knowledge base
Screens, objects, journeys, rules, data and run history. Every run enriches it.
Nothing self-commits
It proposes, a person reviews and merges, and execution replays reviewed code.
How it works
Requirement to report, in five stages.
Five stages read from and write back to one knowledge base. The engine is headless, so editors, CI and ticketing tools are thin clients over the same pipeline.
Scroll sideways to see the full diagram →
The five stages
Requirement analysis
- In
- BRD, PRD, DRS, user stories, Jira epics, existing manual cases
- Does
- Parses and inventories every requirement, assigns IDs, and flags ambiguity and contradiction instead of guessing past them.
- Out
- Requirement inventory with stable IDs
Test case design
- In
- Requirement inventory and the knowledge base
- Does
- Applies risk-based, scenario, boundary and negative techniques against what the application actually does, not only what the document claims.
- Out
- Test design register and traceability matrix
Script generation
- In
- Design register and observed objects from the knowledge base
- Does
- Emits to the framework you choose, such as Playwright, Selenium, Katalon or Appium. Scripts reference real controls, so they do not fail on locators that never existed.
- Out
- Specs, page objects, fixtures, test data
Execution on web and app
- In
- Reviewed, merged scripts
- Does
- Runs against the live product, web and mobile, through the runner you already use. A failure routes back to generation with the observed behaviour attached.
- Out
- Results, traces, screenshots, run evidence
Reporting
- In
- Run results across the suite
- Does
- Publishes results and coverage and raises defects, including requirement defects where the application and the specification disagree.
- Out
- Reports, coverage export, Jira and DevRev tickets
Built to swap
Any model
One provider adapter. Claude, GPT, Gemini, a local Llama, or anything OpenAI-compatible. Swapping one is a configuration change.
Any framework
The design register is framework-neutral. Only the emitter changes, and no runner is the default.
Any substrate
AWS, OCI, on-premises, or a laptop with a device attached. Same engine, same knowledge base.
No model in the unattended run
Generation proposes, a human reviews and merges, and execution replays reviewed code.
Indicative architecture, not a specification.
New · AI‑Lens Visual
Check the build against the design.
Give AI‑Lens a mobile test build and the approved Figma file. It installs the app, signs in, walks every designed screen and compares each element with its Figma layer: colour, size, corners, font and text. The report shows Figma and app side by side with the measured values, separates real defects from differences that need a decision. It connects to Jira both ways: it raises the bugs it finds without duplicates, and reads the tickets designers raise, checks the build against them and posts the result back to the same ticket.
Explored products
Real reports, from real walks.
Each card opens the combined report, where the Knowledge Base tab holds the interactive graph.
Web
Demo Bank
bank-demo-woad.vercel.appABC Demo
bank-demo-woad.vercel.appPinacle
pinacle-frontend.vercel.app/loginQKDemo
qkdemo.inSauce Demo
www.saucedemo.comMobile
See it explore a product of your choice.
Paste a URL and watch AI‑Lens build a sample report. This demo is simulated, so nothing is sent to the site you enter.