No one canshow me how
Small-business owners hear plenty about how AI could transform their companies. The better place to start is one real job they want gone.
A customer told me a couple of weeks ago:
“I go to a lot of these business mixers and they're always telling me how AI can change my business but no one can show me how.”
Jason Freedman recently described a startup pitch that missed for the same reason. After seeing the startup's interface, AI workflow, and assistant, he wrote:
“I’m not interested in making a big shift to how I do things. I don’t care about new UIs. I don’t want to be transformed.
I just want to pay for this particular ask on my to do list to go away.”
He was willing to pay the full services price if the startup could do the work as well or better, and faster, while using little of his time. The startup was charging like a typical B2B software product and still asking the customer to carry the load: adopt the tool, learn the interface, manage the process, and inspect the result.
He was looking for more value than just a login to a series of tables, charts, and workflows. He just wanted the problem to go away.
Nicolai Moelholm commented and pointed out the mismatch, which I appreciated:
Founder: “Our AI will transform how you work.”
Customer: “I was actually hoping to never think about this task again.”
Paul Graham described the founder's side of this problem in Schlep Blindness:
“A company is defined by the schleps it will undertake.”
Stripe is his example. Fixing online payments meant dealing with banks, fraud, security, and regulation. Those were not side issues that appeared after the software was finished. They were part of the problem, and they frightened other founders away.
AI startups can develop their own version of schlep blindness. Building an assistant and a self-serve workflow keeps the customer's bad records, exceptions, existing systems, and judgment at a distance. Taking responsibility for the job brings all of that into the product.
Graham described how this worked at Viaweb in Do Things that Don't Scale:
“Some startups could be entirely manual at first. If you can find someone with a problem that needs solving and you can solve it manually, go ahead and do that for as long as you can, and then gradually automate the bottlenecks.”
He made the same point again recently. Asked how founders can build huge companies without being overconfident, he answered: “Incrementally.” He followed with:
“You put your head down and work. You focus on solving one problem after another. Gains accrue exponentially. And then you lift your head up a few years later and look around, and you're amazed at what you've built.”
That seems like the practical path for AI inside small businesses. Start with one job the owner wants gone. Do the schleppy work. Learn what the business actually requires. Keep what the job teaches. Then earn the next job. Which in the age of agents, I believe will actually help scale this work better than previous attempts in the B2B SaaS world.
The crazy thing to think about: a business may change after enough of those jobs disappear. AI agents can make this possible now.