Having used serverless a bit, I’ve run into many of the same issues and generally agree with the advice but depending on your use case it may not be worth contorting your architecture to fit serverless. Instead, I’d look at it as a set of criteria for if serverless would fit your use case.

At this point in time the only places I’d use lambda are for low-volume services with few dependencies where I don’t particularly care about p99 latency, or DevOps-related internal triggered jobs.