RSI for the Enterprise

Recursive Self-Improvement is coming for the enterprise and it will be one of the most disruptive, autonomous implications the business world, and regular citizens, have experienced in their lifetimes.

I’ve been thinking about RSI for over a decade. We call it Biologic Intelligence, because, well, that’s what it is. Animals recursively self improve all the time. So do humans. Hopefully, anyways.

You make a mistake, you learn from it. Something works, you build on it.

Same thing goes for work environments, what the fanciest among us call “The Enterprise”.

Companies (aka The Enterprise) across the land have two primary resources: people and machines (e.g., software, manufacturing equipment, etc).

The biggest expense in any organization is The People. Owners put up with The People because without it, there would be no Profit. No people, no profit.

People get good at doing the mind-numbing same process steps day in and day out. That’s what the first wave of AI is taking over. Agentic workflows is the fancy term. Basically, automate the stuff that is step by step.

We still need The People to handle the messy 20% that doesn’t fit into those nice, neat workflows.

But Recursive Self Improvement (RSI) is set to take over that last 20% of messiness.

It can adapt and learn on the fly. It can leave notes for itself, ask for help, update its findings and the steps needed to get to the answer or complete the task, and then write down that information for next time. Biologic Intelligence takes it a step further and collapses the logic and the database, making it learn and adapt on the fly even faster.

In this way, it automates the automation process. Thereby removing more work that humans had to do.

We already see this showing up in enterprise productivity numbers reported to investors. Instead of hiring more people, the same number of people can get more work done and improve metrics. Agentic coding is a good example, but so are standard work steps like processing thousands of invoices or responding to basic one-line support requests.

Once The Enterprise begins automating itself, we need less people. That’s good for profits. Owners want profits. The value of their firm goes up the more profitable it is, and its valuation multiple expands the more markets it goes after and captures.

In this way, we can remove people costs from the income statement. We can also remove expensive third-party software costs when we can have our own Enterprise Intelligence Algorithm (EIA?? lol) attack the problems and generate the code and data it needs on the fly.

As we’ve been saying for years, the end game for this whole thing is real-time algorithmic trading at planetary, internet scale.

Those with high-performance memory, compute, and next-gen adaptable algorithms win the game of the future.

Today’s static software with slow cycle times can’t compete against dynamic software with fast cycle times.

If you’re still reviewing PRs manually, you’re not quite getting it.

Of course, many people will say it’s a hoax, or speculative. That’s fine. At some point, the future knocks on your friend door. And you’re either ready to receive it, or completely unprepared.

We’ve been preparing for 33 years. We’ll be fine (bartending on a beach, remember), but the disruption that’s coming is something even the strategy consultants and AI folks don’t deeply feel or understand. They’ve only been thinking about it for years. How could they.

We’re checking out for awhile. You’re on your own.

Good luck.

—Sean