The article does point out numerous challenges with building compilers, and, by extension, with software language engineering, which concerns itself not just with building compilers but with the design of languages, and implementation of non-compiler tools around it. It merely points out the existence, and likely occurrence, of those challenges, and it can be surmised that the author is frustrated by frequent experience in his own daily life.
A bespoke software language -or DSL, if you will- is not always the best solution. It really depends on what you carve out as the domain, how fast that domain changes, what costs and risks are associated with those changes, and with which people you have to implement those changes. No amount of tooling is going to help you out if the domain you chose to recognize as such isn't changing fast enough, isn't costly or risky enough to change (quickly), or the domain stakeholders are anyway not helping out.
But in case you do want to build a DSL, have a look at a language workbench like JetBrains' MPS, or this book I happen to be writing: https://www.manning.com/books/domain-specific-languages-made...