The Mid-market AI Thesis

I have been meaning to start a newsletter, just to continue the dialog with people I have met and interacted with deeply or fleetingly over the last year. But, I hate newsletters. They are robotic. It’s not my style.

So here is a journal of what I think about the AI Zeitgeist as we build Moative.

If you don’t recall from our interaction before, Moative is 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.

So, there is never a dull day or dearth of experiences. I promise to share as we build.

What I also promise:

1. No shilling

2. Opinionated takes – Authenticity over conformity

Who am I?

I will resist a philosophical answer. At work, I am a builder. I have built two tech companies before and sold them – the last was an AI company I built along with Shrikanth (who is the co-founder of Moative as well). This time around, we are building around a few theses. It’s good to start the first letter with that point of view, right?

Mid-market firms’ time has come

AI is a general purpose technology, like electricity or the internet. Unlike SaaS or monolithic ERPs, which are just tools for doing a job, AI redefines what a job is. Steam engine changed movement. Internet changed communication. AI is changing work. 

ERP? Changed the user’s mood from worse to horrendous. 

SaaS? It was like a mug of beer but full of beer froth

It is also why most leaders from traditional industries have a disdain for software tech. The total factor productivity (productivity gain for every dollar spent) for software has remained flat since the 70s.

This is the mid-market opportunity.

Enterprises will be slow to adopt AI. They will

1. Evaluate AI like they evaluated ERP. Software-first evaluation

2. Wait for Accenture and Microsoft to say that it is sage

3. Look for validation from Forrester that it is working

AI evaluation needs process-first thinking. What processes are followed and why? How many of the workflows are built around human skill, departmental, political constraints? You start from there, design an ideal process with no constraints (like ‘Peter is a treasury guy and he won’t understand sales pipeline’), stack rank them by customer value, cost take out potential, and any other levers that matter. Then overlap that with what’s possible today and in the next 18 months with AI. You get a roadmap. Pick the low-hanging ideas where you cannot go wrong.

AI is evolving but not evenly. Some parts like document context extraction, transcriptions, simple binary decisions, etc. are all solved for most parts. 

What if you don’t need an ERP and instead a swarm of agents and a home-grown system that needs minimal human intervention? Imagine the freed up time and dollars.

What if level 1 and level 2 customer support tickets are automated?

What if all of these can be incrementally rolled out in monthly increments?

The cost of writing code is falling down exponentially. So it is crunch time for packaged software companies, but the exact opposite for a midmarket CEO who is willing to grab the opportunity to reduce the cost base and attempt growth experiments.

Here’s the economics of AI failure that no one talks about: A failed ERP implementation sets you back by 2 years and several million dollars. A failed AI roll out, sets you back 45 days and tens of thousands of dollars.

Anyone can and will build AI solutions. In the consumer world, what is left to discover is taste (clean, delightful, and efficient interfaces is an example). In the industrial world, its data and context. That does not come from your vendor. It comes from you.

This is our first thesis.

This AI platform shift has elevated the value of data and the context embedded in the processes. With falling costs of software development, a midmarket company can wrangle software to fit their world view – without spooking their board or staking their financial stability.

So if this thesis is right, how does Moative approach it?

With open minds and a clean slate. No software to sell. No revenue to protect. We work backward from the opportunity that the board believes in. We then get to work. But, if we like the opportunity we put our skin in the game.

If AI is so powerful and can change industries, are we willing to bet on it? Yes.

How? We form joint ventures with some of our customers/partners that are experts in their industries, to build out a future we both believe in. The business model follows.

I will follow-through next week with the second (and the last) thesis of ours. After that, it will be us, reporting from the trenches.

Enjoy your Sunday!