Ten thousand dollar productivity > Million dollar transformation

"AI is changing everything" is a tiring trope, if we are not able to show concrete examples of how it changes daily work. Companies that want to open their purse strings, want to know how AI can make a difference in days. 

Consulting firms see AI as a "transformation" opportunity. Analytics firms see it as a "data unlock" opportunity. Data tool vendors see it as an "orchestrate for efficiency" problem. None of these are intuitive to the buyer – the ultimate one who carries the wallet.

We are setting up a new product team. Intuitively we know AI is changing everything. But what does it mean for this greenfield team? We started with our comfort zone, threw in a couple of engineers, a quality assurance analyst, a couple of team leads for engineering and GTM. We already got to 10 people. Zero code written. Then I looked up 'application layer companies' in AI. The best ones had $300K revenue per employee. Foundational model companies had $1M per employee.

We look back at our own consulting firm. We are somewhere between $100K and $150K. In other words, we are better than where SaaS is today. And yet, when it comes to building a product startup, we are going back to our comfort zone. We equate people to velocity. So we dumped the bottom up rationalization and decided that we will hit at least $2M in revenue with 10 people. That is $200K per employee. Given the trajectory of AI and the maturity of software productivity tools, we are probably under-estimating. We should push for $300K per employee.

While we were doing this exercise, a couple of us upgraded to the 'Plus' version of Claude Code and built internal tools for hiring, project management, GTM – all things that would have cost us $1500 per month and now replaced with code written over 10 days and $200.

We realize that taste is the limiting factor and not tools or people. If you have a good point of view, any software can be built fast. To build a good product in the early stages, you need to think more like an engineering craftsman and less like a CTO. If you are not a software company and yet have a point of view for what software should do for your process, you don't need a CTO or a tech team. AI has democratized software-led productivity.

Now look at your industries and your company if you are not from tech. According to Epoch.ai, the primary contributor to labor productivity does NOT come from R&D and it comes from general labor workforce productivity. And yet, most of the visionary AI leaders talk about how AI will cure cancer. Yes it might, but that is not the path through which the technology will diffuse into becoming salient at work. It will be through freight automation instead.

Automating mundane things will unlock more than building weather models for crop resilience.

This is where companies of all sizes should rethink AI investments. Data harmonization is important, flow of context across the org is important even to unlock labor productivity. But ditching an applicant tracking system and building your own light-version that saves time for your HR team is just as productive and valuable. Big bang transformation and small and quick wins should not be mutually exclusive.

If every company is going to become a software company, if you leave out the jobs for which physical presence is required or abstract reasoning is important, what you are left with are roles that are common across industries. What would it take to remove old-school software that pulls back the productivity of these functions? What can you custom-write that will increase labor productivity in these roles?

If you ask an analytics or consulting firm they'd say that by doing these you'd create point solutions that silo the data. I'd say, "So be it!" It's no better than having an employee who does a bunch of work on emails, phones, and spreadsheets because the current crop of software has failed her. There is labor unlock through automation that is possible today without a single dollar invested in Databricks, Snowflake, or Accenture.

If you are looking to embrace AI and feel the real progress with no downside risk, set aside $10000 per month and automate the mundane relentlessly. It works.

Here's our offer: Give us an hour and 10 mundane things your best employees are burdened with. We will knock off one of those every month.

Enjoy your Sunday!

Keep reading