I had a vendor that made a special RS232 to multi-point printer protocol translator. It was slow and unreliable. I build a small C program that did the job from inside the UNIX machine. Bye bye vendor. The C program logged the data to the hard disk in the machine, so we could "replay" all the traffic if necessary. A capability the dedicated device could never do.
Lots of other big money savers inside the PacBell machine room. Suffice to say, my group returned a savings of $6 for every $1 expended on our salaries.
They avoided hiring well over 250,000 people at $40,000 per employee per year, or $10,000,000,000 perannum for an average of 17.5 years. That is a huge amount of money ~ $175,000,000,000 ~ worth just a bit more than two full years of total postal operating costs.
Frankly, I'm still mildly upset about not having gotten a special performance award for the effort ~ or maybe promotion to the top job at the age of 23 or so.
One of the problems you have with government work is that most folks involved in it actually imagine that the reason your idea was successful is that you were simply in the right place ~ not that you were a bright person who worked hard.
One of the guys who several years later created some useful enhancements to the basic device did end up as Deputy Postmaster General ~ and that was the end of his career. Bright guy; creative; but the Board of Governors has never approved of either attribute.