Eggplant Functional Testing: Complete Overview of Keysight's AI-Driven Platform

Eggplant (now Keysight Eggplant) is one of the oldest and most distinctive test automation platforms on the market. Unlike most testing tools that interact with an application's DOM or API layer, Eggplant uses image-based, model-driven testing — it looks at your application the way a human does, through the screen.

That sounds clever in theory. In practice, it means Eggplant can test things that other tools can't: embedded systems, legacy mainframe UIs, ATMs, set-top boxes, point-of-sale terminals. If there's a screen, Eggplant can test it.

This guide covers what Eggplant is, how it works, what it costs, and where it makes sense — and where it doesn't.

What Is Eggplant?

Eggplant was originally developed by Testplant in Scotland, acquired by Eggplant Software, and then acquired by Keysight Technologies in 2020. It's now marketed as Keysight Eggplant or just Eggplant DAI (Digital Automation Intelligence).

The platform has two core components:

• Eggplant Functional — the legacy scripting engine, SenseTalk-based, connects to a device via VNC and drives the UI through image recognition and OCR
• Eggplant DAI — the modern AI-driven layer, adds model-based testing, analytics, test case generation, and coverage optimization

Most enterprise customers use both together. DAI generates and manages the test model; Functional executes the tests against real or virtual devices.

How Eggplant Works

The core mechanic: Eggplant connects to a System Under Test (SUT) via VNC (or its own proprietary protocols for some device types). It captures a live screenshot, finds UI elements by comparing against image templates or using OCR, then sends synthetic mouse clicks, keyboard input, or touch events.

No DOM access. No JavaScript injection. No API hooks. Just pixels.

This makes Eggplant technology-agnostic. It doesn't care what framework your app is built with. It doesn't care if you have source code access. If it runs and renders a screen, Eggplant can test it.

SenseTalk Scripting

Eggplant Functional's scripting language is SenseTalk, a plain-English-style language that reads almost like pseudocode:

Click "Login Button"
TypeText "user@example.com"
Click "Password Field"
TypeText "securepassword"
Click "Submit"
WaitFor "Dashboard"

The strings like "Login Button" refer to image templates stored in your test suite. When the script runs Click "Login Button", Eggplant searches the screen for an image that matches the stored template and clicks the center of it.

For text-based elements, Eggplant uses OCR so you can reference text directly without storing images.

Model-Based Testing in DAI

Eggplant DAI shifts the paradigm. Instead of writing scripts, you build a behavioral model of your application:

1. Define states (screens/pages your app can be in)
2. Define actions (transitions between states)
3. Define coverage goals (what paths need to be exercised)

DAI then automatically generates test cases that achieve the coverage you need, using AI to prioritize paths and identify gaps. It can analyze historical test data to predict which areas are most likely to have defects.

Architecture and Deployment

Eggplant supports several deployment models:

On-premises: Full installation on your infrastructure. Eggplant Functional runs on Windows, macOS, and Linux. The DAI server is typically a Linux-based deployment, often containerized.

Cloud: Keysight offers SaaS hosting for DAI. Eggplant Functional runners can be cloud-hosted or on-prem, connecting to cloud-hosted SUTs.

Hybrid: DAI server in cloud, execution runners on-premises (common for testing devices that can't leave the network).

License Model

Eggplant uses a concurrent user licensing model for Eggplant Functional. Licenses are expensive — typically $15,000–$40,000+ per seat annually for enterprise deployments, though exact pricing is negotiated and rarely published.

DAI adds significant cost on top of Functional. Total platform cost for an enterprise team often runs $100,000–$500,000+ per year.

This pricing reflects Eggplant's target market: large enterprises in highly regulated industries where the cost of testing failures (banking systems going down, medical devices behaving incorrectly, telecom infrastructure failing) dwarfs the tool cost.

What Eggplant Is Good At

Legacy and non-web UIs: Mainframe terminals, thick-client Windows applications, embedded systems — anything that other modern testing tools can't touch.

Cross-platform consistency: Because Eggplant sees the screen, it treats all platforms equally. The same test can run against a Windows app, macOS app, iOS app, Android app, or an ATM — as long as you can connect via VNC or the appropriate protocol.

Accessibility testing: Eggplant's image-based approach naturally covers visual layout. You can verify that UI elements appear where they should, are the right size, and are properly labeled.

Regulated industries: Banking, healthcare, telecom, and government organizations use Eggplant because it provides exhaustive audit trails and can test systems they can't modify or instrument.

Model-based coverage: DAI's model-based approach helps teams achieve systematic coverage of complex state machines — useful for workflows with many possible paths.

What Eggplant Struggles With

Web applications: For standard web apps, Eggplant is overkill and often slower and harder to maintain than Playwright, Cypress, or Selenium. Image-based testing breaks when UI elements resize or reposition due to responsive design.

Speed: Image recognition is slower than DOM interaction. Test suites that would run in minutes with Playwright can take hours with Eggplant.

Maintenance overhead: Image templates need to be updated every time the UI changes. In fast-moving web applications, this creates significant maintenance burden.

Cost: The licensing model puts Eggplant out of reach for small teams and startups. The ROI calculation only works at scale in high-consequence industries.

Learning curve: SenseTalk is approachable but the full platform — model building, device management, analytics configuration — requires significant training investment.

Eggplant vs. Modern Web Testing Tools

For web applications, Playwright or Cypress will outperform Eggplant on every dimension except legacy system support.

Who Actually Uses Eggplant

Eggplant's customer base skews toward:

• Banks and financial services testing trading platforms, ATMs, and mainframe applications
• Healthcare organizations testing medical device UIs
• Telecom companies testing network management interfaces
• Government agencies testing legacy systems
• Defense and aerospace testing embedded system UIs
• Retailers testing point-of-sale terminals

If you're building a modern web application or mobile app, Eggplant is almost certainly not the right choice. If you're responsible for testing a mixed estate that includes legacy systems alongside modern apps, Eggplant's unique capabilities may justify its cost.

The AI/ML Layer

Keysight has invested heavily in positioning Eggplant as an AI-driven platform. The DAI component's AI capabilities include:

Test generation: AI analyzes the behavioral model and generates test cases that maximize coverage with minimum test count.

Defect prediction: Machine learning models trained on historical test results predict which parts of the application are most likely to have defects.

Self-healing: When image templates fail to match (due to minor UI changes), the AI attempts to find the nearest matching element.

Analytics: DAI provides dashboards showing coverage trends, test efficiency, and risk areas.

These AI features are genuinely useful in the context of large, complex test estates. They help teams prioritize what to test and reduce wasted effort on tests that consistently pass.

Integration Capabilities

Eggplant integrates with the standard enterprise testing stack:

• CI/CD: Jenkins, Bamboo, Azure DevOps, GitLab CI
• Test management: Jira (via Xray), qTest, Azure Test Plans
• Performance testing: Keysight's own performance tools
• Reporting: Built-in analytics, plus export to external BI tools

The integrations are functional but not as smooth as tools built natively for modern CI/CD workflows. Configuration is more complex and often requires custom scripting.

Getting Started with Eggplant

For enterprise evaluation:

1. Request a demo from Keysight — they don't have self-service trials for the full platform
2. Identify your SUTs — map out which systems you need to test and confirm they can be connected via VNC or supported protocols
3. Start with Eggplant Functional — get a working script against one application before investing in DAI
4. Build incrementally — the model-based approach in DAI requires investment; start with manual scripts and migrate as you understand the platform

A Simpler Alternative for Web Testing

If your team primarily tests web applications, Eggplant is solving a problem you don't have. Modern tools like HelpMeTest let you create browser tests in plain English without writing code — no image templates, no VNC connections, no $40,000 license.

HelpMeTest runs on top of Playwright with AI-powered test generation, self-healing tests, and 24/7 monitoring — with usage-based pricing at $0.003 per test run and no base fee. For most web development teams, that's the better trade-off.

Summary

Eggplant (Keysight Eggplant) is a specialized testing platform that genuinely excels in scenarios no other tool handles well: legacy UIs, embedded systems, and any application where you can only interact through the screen. Its AI-driven DAI layer adds meaningful value for large test estates in complex enterprises.

For most teams building modern web or mobile applications, Eggplant is the wrong tool. The cost, complexity, and maintenance overhead aren't justified when Playwright, Cypress, or AI-powered tools like HelpMeTest can do the job better, faster, and at a fraction of the price.

Know your context. If you have ATMs to test, call Keysight. If you have web apps to test, pick something built for the modern web.