The Big Shakeup
We are all living through an incredible and tumultuous time in history. Age old institutions are being shaken up as the ways of conducting business change completely under our feet. AI is currently being cited as an excuse by huge businesses like Amazon, Microsoft and even Target to make cuts to their spending. Many employees who paid the cost of these savings are venting their frustration online at how the economy is reeling under the pressure of a powerful new technology that seems to have suddenly found itself upon us. The problem these firms will quickly face is obvious to anyone who has been hands on with AI for any period of time, that while the technology is very powerful, it is still not mature enough to replace human operated systems unilaterally. If the story ended there it would be great news for workers who have expertise unique to their industries, but unfortunately, this technology’s rollout is only just beginning.
What is an AI Workhorse?
Enter 2026 and in just the past few months we’ve seen massive improvements in the world of AI research and products. New models, better models, new workflows and some novel use cases. There is a service called Clawd (not to be confused with Claude) which allows its users to spin up an “LLM agent” on a machine of their choice and take full control over their PC, execute code, whatever it wants! The users even let it have access to their emails, personal documents, passwords in an effort to automate their busywork. They’ll often provide it with vague tasks to complete without and allow the bots to attempt to “think” through the problem, often burning through their tokens in the process.
I think it’s appropriate in the Chinese zodiac year of the horse that people are creating these new “AI Workhorses.” Like horses, these agents work hard at any task you set them to and will stubbornly apply themselves to it until it is done or they run out of energy. What I’ve found in my work with AI is that like these workhorses, the more care you take of them the better they become at working with you. You learn their personalities, what they’re good at and train them on new tasks that they can become experts in. While your results may vary trying to employ AI to do any particular thing, people who know how to use these beasts to enhance their own workflows and thinking will become unstoppable and will be armed with technology that they can ride off with anywhere their work takes them.
Until we have AGI, defined as an AI capable of performing “any” cognitive task better than a human being, each AI implementation needs to be treated as its own creature which will behave and perform according to how it was raised and created. Models that are trained on human sounding output perform better at those tasks than models that are created for coding. Individual conversational models can be refined to speak in particular ways or even solve particular cognitive problems for their users. Knowing how you intend to apply an AI in your domain is the first step, but understanding how to scrutinize them takes hands-on experience. only it was so easy that you could just look them in the mouth.
Where Does that Leave Us?
Today, unless you’re working in software or hardware development, the technology is evolving too quickly to keep up with. It’s not necessarily safe to allow your employees to run AI through their emails and help them communicate with clients, even though it may help them significantly. These tools should be deployed sandboxed and secrets should be kept off the internet, away from any LLM deployment which might have access to them. Many believe that the rules of business have completely changed, but they have not. For now, if you want the most out of AI without waiting for big tech companies to answer your prayers, look for professionals who have already solved problems for themselves using these tools and replicate them in your own business. Look at all the software that your business is built on and keep an eye out for companies that can provide new versions of similar services. It’s going to be small companies like Fontaine Tech who are best positioned to solve real problems in their communities using software.