Describe the problem in ordinary language
What are you trying to do, and what makes it difficult today? Describe the task before listing possible technologies.
Check the fit
Consider the input you have, the output you need, and where the result will go next. An impressive demo is not evidence that a product fits your task.
Look at the whole cost
A new tool may need setup, ongoing attention, data access, and training. Consider those requirements alongside its subscription or purchase price.
Keep an exit
Check whether you can export your work and whether another person can understand the process. A useful tool should support your work without making it unnecessarily dependent on one interface.