I like optimizing things, how do I make this into a business?

mid-30s male, IT career, retired. But I'm bored. The thing I'm really good at is optimizing things, streamlining them, or creating new tools to make life easier. For example:

• I worked for a major storage vendor, and one of their bigger customers had a problem where the storage array would crash and stop servicing IO requests when it shouldn't have. I was given a copy of their hardware setup (as a junior role, by myself; originally it was supposed to be a principle + 3 seniors to help him) and found the issues at play. Bugs in various vendor drivers and the storage array team not using up to date drivers in their own product. To do that, I had written my own IO load tester that could scale to 10s of thousands of machines. I built an architecture that would let me start/stop the ecosystem in about 5 minutes, and I could rebuild the entire thing 'from factory' in around 4 hours. For reference, for professional services to do the same task took 7 working days.
• I visit the same BBQ place every weekend and hang out. I had them change their opening procedure, resulting in tasks getting done first thing in the morning, rather than during a busy lunch rush. I set them up with a nice sliding magnetic sticker so the person at the window could glance back and see how much of each meat/potatoes/sides/etc were left. I talked with some of the underperforming employees and made them reasonably useful. The owner pulled me aside and said I was worth about 30% of his revenue at this point with all my tweaks. He was considering closing shop 8 months ago, now he's considering being open more days.
• I worked at an IT lab that had both ethernet and FC patch panels. 10,000+ ports. Constant churn where endpoints on both sides would change, as well as patching changes. I created a (very simple) web page where you could give a source and target WWN (or MAC), and it would give you 1) a list of instructions on how to physically move patch ports around (what from where to where) 2) Give you the zoning commands to clean out the old/existing zones 3) Give you the zoning commands to create new zones specific to the switches being used 4) give you a 'confirm' button to click once you'd made those changes to mark them as 'done' rather than 'pending'. This saved hundreds of hours.
• Another company I worked at got support requests from customers. Customers would supply the URL of their instance (think cloud web hosting) and to resolve that to a set of physical machines running their instance took about 15 minutes, and the use of 4 different internal tools. I wrote some stuff to pull/crawl/parse those tools and put it all in one central location, allowing you to paste the customer URL in and instantly get the machines, load balancers, zones, regions, etc. All on one page. Each query saved, on average, 15 minutes, and in the first year it saved around 10,000 employee hours (it was used by a BIG company).
• Sometimes documentation is the answer. At one place I sat down and converted tribal knowledge + partial documentation into an ansible script that took a 2 week process down to 20 minutes of user interaction and about 6 hours of script time.

So…this is an example of the sort of problems I enjoy solving. I've read through the various consulting pitches and, well, I don't really see anyone trying to sell consulting in the form of "hey, your internal structure is inefficient and I can make it better." or "Let me be the person who writes down all the tribal knowledge." and it doesn't sound super profitable, but in big companies it can be hugely profitable.

I think where I get lost is that big companies (Oracle, Google, Kroger, etc) all expect to hire that 1 in a million employee who will elevate things to greatness before deciding the pay sucks and leaving. And if I'm going after smaller companies, I'm going to spend the same time making fixes to their process, but the gains are going to be far smaller.

I've not really seen a market need for any of the tools I created. The ansible script? Only that company would ever use it. The patch panel software? Yeah, probably, but the main thing people would want is a way to log in and scan endpoints to update the DB automatically, and that's a HUGE security problem.

I guess, is there a niche or term for 'let me come in and just make things better'? I don't have a solution I can pitch because I figure out what to fix once I'm inside. And, frequently, what I identify as a fix is NOT what management sees as a problem; the best tools I've created were all done in spite of management forbidding me from working on them. I knew mgmt was wrong and ignored them. I am very much a 'disagreeable' personality, though, I've mellowed with years.

Final thought: What's fair compensation? Or, maybe standard? If I save someone $20m/year, recurring, I feel like at least $10m is fair. If I save you $20m once, I feel like $2m is reasonable. Am I way off base? Too high/low?