We're building AI to replace customer service jobs.
People are going to lose their jobs to tools like this one. Not moved into higher-value work. Replaced entirely.
If too many jobs get replaced without replacing the salaries, the world has a very big problem on its hands. We don't have all of it figured out yet. We're starting by giving back to the people this technology replaces.
Automate away the workers, and you automate away the customers.
Here's the part the spreadsheets leave out. The worker a company just replaced was also somebody's customer. Do that across enough of the economy and you don't get a lean, efficient future. You get one where almost nobody has a paycheck left to spend.
Everyone who earns a paycheck also spends one.
And universal basic income doesn't quietly ride to the rescue either. UBI is paid for out of tax dollars, and the two biggest sources of those, income taxes and payroll taxes, are built mostly on wages. No jobs, no wages, no tax base, no UBI. The safety net is paid for by the very wages all this would be cutting.
So handing some of the value back isn't charity, and it isn't guilt. It's the only version of this future that doesn't fall in on itself. Automate the whole world out of work and you've automated away the people who were going to buy what you're selling.
If it learned the job from you, you should own a piece of it.
AI didn't come from nowhere. It learned from the work of nearly everyone who came before us, from the answers we wrote down to the problems we quietly solved. It is more like an inheritance that everyone built than the private property of the few companies that trained on it.
So the wealth it creates should find its way back to the people it came from. That won't happen on its own, and it certainly won't happen if AI companies do what companies usually do, pocket the savings and move along.
Celeste is a small, honest version of the same thing. The agent learns your business by reading your team's real work, and the day it takes a job over, it is building on the work of whoever did that job first. We think that person should own a slice of whatever it goes on to earn.
How can you even afford this, and why do it in the first place?
It's not philanthropy. It's math.
AI is driving down the cost of labor, fast and hard. Like most technology companies, we've begun using it across our own business, and the savings are large. That is a big part of how we can afford to give most of our profit away.
The easy thing is to keep those savings, the way almost every company will. We think that is short-sighted, and short-lived. If the gains from automating work keep concentrating in the accounts of a handful of AI providers, the economy runs into serious problems, and fast.
Until a better plan comes along, the only real solution we can see is to share the profit this technology creates, widely enough to keep the economy going. So we give ours back. Not out of guilt, and not to look good.
It's not philanthropy. It's math.
Here is what we do.
Every year, we set money aside for the workers our AI replaces. It is the larger of two amounts: 5% of what we take in, or 70% of our profit. They get it as a share they own, not a one-time check. Because one of the amounts comes from sales, it starts with our first dollar.
Our goal is to give them a real stake in the company. This is already in force. Today it runs on a form and a checkbook. Later, a trust will hold it for good.
Our first pledge is simple. Bring us a customer, and you keep 70% of the profit on that customer, every month, for as long as they stay.
If we want things to be different, we have to do things differently.
What we won't pretend
The mechanism isn't finished, and we can't promise nobody gets displaced or that we'll make anyone whole. In a year with no profit there's nothing to share, and we expect our first few years to be lean. Better you hear that from us now than later.
Why there are no investors
We're bootstrapped, on purpose. A promise to send most of the value away from shareholders is hard to keep when the company is owned by people expecting the opposite. By the usual measures, that makes us a poor investment. We wrote that down too. Read the investor note.