AutomateAI  /  Platform
Platform

Five agents. One way of reasoning about your product.

Every AutomateAI agent works the same way: look at the screen, decide what a person would do next, act, and check the result. The only thing that changes is what kind of screen it's looking at.

Web

Drives real browser sessions like a person would

The web agent reads the actual rendered page — not a fixed selector map — and decides what to click, fill, or wait for based on the goal it's given.

  • Resolves ambiguous elements explicitly instead of guessing and reporting a false pass
  • Recovers from overlays, redirects, and layout shifts mid-run
  • Distinguishes a genuine navigation from a click that silently did nothing
  • Captures network activity and console state alongside every run for debugging
Recent selector decision
CHECK 3 elements match "Men" — narrowing
OK Scoped to nav link, not category tag
CLICK Men — Shop all
Mobile

Tests on real Android and iOS devices

Runs against your own device farm, tracking app state through auth, deep links, and the kind of flaky navigation that only shows up on a real handset.

  • Understands native and Flutter-rendered UI alike
  • Coordinate-aware, so taps land correctly regardless of screen scaling
  • Loop and stall detection so a stuck screen fails fast instead of hanging
  • Works from an installed build or an uploaded APK/XAPK directly
Device session
TAP "Continue" — onboarding, screen 2/3
WAIT Screen settled — 340ms
OK Deep link opened correct screen
Chatbot

Holds a real conversation with your bot

Instead of sending isolated messages, the chatbot agent carries context across turns the way an actual user would — then checks whether the conversation stayed correct and on-track.

  • Multi-turn conversations with a defined goal, not single-shot prompts
  • Checks whether responses stay factually and contextually consistent
  • Flags when the bot drops context or contradicts itself mid-conversation
Conversation check
TURN 3 Asked about order status
OK Bot correctly referenced order from turn 1
TURN 5 Checking tone stayed on-brand
Accessibility

Audits every screen it already has to look at

Since the agent is reading the rendered screen for every other test anyway, it runs a structural pass, a vision pass, and a cross-check between the two — mapped to the standards your compliance team tracks against.

  • Three-layer audit: structure, vision, and cross-reference between them
  • Findings mapped to EU EAA and US ADA / Section 508
  • Clear pass, fail, and needs-manual-review states — never a vague "not tested"
Guidelines
  • Color contrast (text) PASS
  • Focus order PASS
  • Screen reader labels MANUAL
  • Touch target size PASS
Performance

Times the app the way a user actually feels it

The agent already knows when it tapped something and watches the screen directly, so it can flag real response-time regressions without an APM SDK anywhere in your codebase.

  • Detects when the screen has genuinely settled after an interaction, pixel by pixel
  • Flags regressions against your own historical baseline automatically
  • Report styled to match the rest of your test results
Screen-settle latency
tap detected
How it reasons

Every run is a bounded loop: observe, decide, act, verify

Agents work within a fixed step budget per run, so a confused agent fails cleanly with a reason instead of looping forever or running up cost.

25
max reasoning steps per run
1
forced verdict on the final step — no silent timeouts
5
agent types sharing one execution engine
100%
of runs report a reason, pass or fail