SaaS, contrary to some opinions, propelled the U.S economy through the 2000s and 2010s. But, as some have pointed out, it's actually very hard to copy what makes a company successful and this almost never works in practice.
But maybe this is about to change. AI agents being loosed on the economy are systematizing process knowledge in workflows which previously more or less lived in the heads of employees, or maybe more accurately, in the brain that the employees make up as "the company".
This leads to a situation where both the software the company sells and also the process knowledge and expertise to build that software is fully digitized.
Is there really anything a company does which could not be written down in a .MD file? Of course this could have been done before but then you would still have the problem of how do you slurp down all of that knowledge? No single person would have the time or motivation. But modern AI is highly competent in the text realm and it can process orders of magnitude more text, more quickly and attend to the important points more expertly without being distracted.
Another aspect of this is the circular nature of having made text slinging guru AI's which turn everything into text, which then goes on to effectively represent the immaterial soul like aspects of companies which have eluded copying so well.
Company as software is something new entirely and It's hard to predict where it will lead. Darwinian dynamics may start to ramp now that there's both an external optimizer in terms of the market and essentially DNA that can be acted on. Whereas there was basically no DNA analog before, so the optimization process may have been much less efficient.
When asked what he thought was the most transformative invention in recent history, Sam Altman said that it was the Joint Stock Corporation. In a similar vein, maybe the most important invention, looking back in a few hundred years won't be AI itself, but some byproduct like company as software or some unthought of abstraction.