Thesis Note: Intelligent systems & Interdisciplinary research

Quick Refresh, in case you are new here: I run Moative, a hold co., that creates AI-first businesses. We have Moative Agentry focused on mid-market AI transformation and Leslie (yet to be launched) focused on AI for local governments. We have one more in the pipeline.

We talk about the theses that guide us. In the last essay, I wrote about our mid-market thesis. You can read it here.

Now, let’s talk about our second thesis.

Intelligence beyond software

I will make the case for our second thesis by presenting the anti-thesis (well, not the exact literary device, but you get it).

Software (of the commercial, off-the-shelf kind) is a commodity. I am not saying this because of AI. As someone who could never code, I have a very personal and correct (of course!) view that software has always been a commodity. Vision, strategy, and ideas matter. Commercial software products were a launchpad to execute those strategies. The science and art of software was un-intuitive that getting it right meant something. It was a moat. Good software inflicted less damage on the user. Good software sped up execution. Good software needed good people to write it. Good software became a competitive advantage.

But, in an ideal world, if the idea is in my head and software would exactly write it the way I imagined, the moat shifts from software to the other things like vision/taste, strategy, distribution, personalization, etc.

We are in the world, I wanted 20 years ago.

Software is not yet a commodity. Software is till not easy to get right. The protocols to get AI to write software, integrate with tools, observe if they work – these are all still in the works. But we are directionally heading towards software being easy, if not commodity.

The end of software product and system integrator era

Until now software was rules-based. Intelligence came from the humans. This is why the best of CRM was still dumb enough to ask humans to tell what happened in the call. We all hated CRMs or ERPs or whatever the trauma your company inflicted on you during your formative years.

Now intelligence has moved from biology to code. Language, context, and reasoning are all available on the tap. If you have PhD. level intelligence available on the tap, will you build a CRM or a rocket?

I am not even going to make the argument for what makes for a good venture-fundable company. I don’t care. But the question of what could one build, stands on its own. What problems can you solve for humanity with abundant intelligence on the tap?

Then there is the question of moats. If its easy to build productivity software (the word productivity will take an ironic twist in the next few years), it will be built in spades. Someone with better taste will make AI-led productivity tools that actually make lives of workers delightful and they will win, provided they capitalize their builds and conquests smartly. Well, I promise not to talk about the VC dynamics here. That is for another essay. If software is easy and horizontal productivity workflows can be quickly and efficiently automated, the moat isn’t in building it. It is in distributing it. Besides, if its all easy the user expectations will increase. They won’t want your software. They would want theirs – the software the truly is built around their realities. I think its a good time to be an orchestrator of custom-configured intelligence for businesses. I won’t call them system integrators. That is an inadequate word in this era.

The beginning of intelligent everything

The cutover from software to devices has been happening for a long time. The industrial world has unimaginative names like Industry 5.0, outcompeting the software world in blandness. Be that as it may, intelligence on any and all devices is a mega-trend. Intelligence to solve fundamental and ‘within the reach’ bottlenecks in foundational disciplines, applied disciplines and industrial use cases is now available.

The moat is shifting from software craft to inter-disciplinary problem solving (Physics for robotics, bio-chemistry for drugs, etc.). Research inefficiencies that stymied the time to market can now be broken through.

Our second thesis

It is untrue that computer science is useless as a college course. Just like languages, intelligence (data science to begin with) will be a foundational course for everyone. Artificial Intelligence will be a necessary subject to understand for every discipline in engineering and science. This will happen over the years as academia catches up.

Until then, bringing inter-disciplinary research teams and AI engineers to envision new things is by itself a moat. Hard challenges, new types of teams, and new types of solutions – this will happen a lot.

We don’t have a play yet here but we are tinkering on the sides by building intelligent sidekicks for biologists. We will talk more about this another day. This is one of those things where you wish it happens and the universe makes it happen.

If you are in hard science, deep tech kind of areas, we will go to great lengths to work with you. So hit me up.

From the coming week, we will shift gears to what we learn as we go. I hope the last ten minutes was worth your while. Let me know.

Enjoy your Sunday!