We schedule all of our tests through DAI Controller. It is the only way to schedule DAI runs (besides using APIs I guess) and so the only way to take advantage of DAI’s “Automation Intelligence” and the only way to run tests created by model test cases or model exploratory tests. If you are just scheduling script-based tests then there is no real benefit over scheduling them in EPF–aside from having your results funneled up to the DAI dashboard and being able to centrally manage your SUTs.
The dashboards kind of suck. You have to filter per model and cannot show a true dashboard of collected results. I would like much more control over what is displayed–and some means of sharing with management besides screenshots (I know, APIs again…).
Everything seems to run a bit …worse… out of DAI. Novice engineers have a heck of a time creating robust enough scripts that their results can be dependable. A test that runs dozens of times in EPF will often fail for wait times and other issues when run out of DAI.
We still see frequent errors out of DAI that require us to reboot our server (we run our Agents on our server VM). About once a week DAI just says it gave up waiting for a response from our Agents. No solution yet.
Everyone must to be a member of the same group in order to see what their Agent is scheduled for. You can’t have sub-groups within an Agent or they would schedule conflicting runs.
DAI is complex and “one more thing” for engineers to learn. To use it effectively engineers must script their tests differently than they might otherwise. Scripts must/should be “modular”.
Honestly we have a bug/feature request a mile long with Eggplant on DAI and EPF issues. But we still use it because DAI’s “automation intelligence” has a lot of potential, and Eggplant is working to make improvements.