Eggplant AI from Testplant: No test automation experience, no problem
At the start of November, Testplant launched Eggplant AI, a brand new addition to the digital test specialist’s Eggplant Digital Automation Intelligence (DAI) suite of products. Following the reveal, NCC Group was invited to participate in Eggplant AI’s early access programme.
You can learn more about the launch event in a recent blog post by Paul Bell here:
Paul is well-versed with automated testing and knew that his first impressions would be influenced by everything he has learned over the years. He set his sights on finding someone without as much experience who could also trial Eggplant AI.
As a test analyst with no previous experience of test automation, I was asked to trial the software and assess how accessible it is to someone from a ‘non-technical’ background.
Getting up and running
Using Eggplant AI requires the user to have Eggplant Functional downloaded, as well as an AI ‘Agent’ to communicate between Eggplant Functional and Eggplant AI. If the system under test (SUT) is a mobile device, the user will also need the relevant ‘gateway’ for the SUT (either Android or iOS). If you plan to access and run the software on a machine that requires admin credentials for downloads or approval for software blocked by group policy, this is worth bearing in mind.
Once I had everything downloaded and set up as per the user guides available from TestPlant, AI was an incredibly easy piece of software to use. In summary, a user maps the features of an SUT into a Model within Eggplant AI, attaches test cases written in Eggplant Functional, then sets the AI system off running through that Model for as many iterations, or runs, as the user choses.
The process is still very reliant on the user, in that I needed to build the Model in AI, write the scripts in Functional and link the two correctly for the system to be able to run. However, the ability of Eggplant AI to choose the paths it takes through the Model, learning as it goes and revisiting paths where it has detected issues or defects certainly goes some way towards justifying TestPlant’s claim of Eggplant AI being a way to automate exploratory testing. It is this element of artificial intelligence and the speed at which multiple runs can be completed that really brings Eggplant AI into its own.
How did I find it?
I found the Eggplant AI web interface intuitive, user friendly and mostly self-explanatory, with more than enough resources and tutorials provided by TestPlant to guide you through the process when needed.
TestPlant’s support team is also excellent, responding quickly to any queries and always readily available for WebEx sessions when requested.
As I had no prior knowledge of scripting using Eggplant Functional’s SenseTalk language, I also had to learn this. But with Testplant’s support and documentation I was able to compile basic scripts without too much trouble. Then, leaning on my experienced colleagues here at NCC Group, I was able to produce a fully functioning Model driven by these scripts within just five days.
All in all, I agree with Paul’s view that some of the claims currently being made by TestPlant are yet to be proven. However, I’m sure that as NCC Group’s partnership with TestPlant develops, we’ll come to know the company’s products and capabilities even better than we currently do and start to fully appreciate the potential disruption the Eggplant DAI suite may cause.
Published date:  29 November 2017
Written by:  Lauren Garner