I named my finance automation agent Nisaba, after the Sumerian goddess of writing and numbers. It felt right: Humanity's first writing wasn't poetry or law, it was accounting—tallies of grain pressed into clay. The scribes who kept those records became some of the most trusted figures in the ancient world. Get the count right and the temple, the harvest, and the whole economy held together.
Five thousand years later, I've automated the counting. Nisaba reconciles, matches, and categorizes faster than I ever could. But I've noticed something in myself: I have never once sent a client a number sourced straight from an agent’s first output. When real money is involved—an invoice, a payment, or a figure someone is relying on—I check every penny myself. Not because the machine is usually wrong, but because I'm the one who has to stand behind it. I can't tell a client, "Sorry, my AI missed that."
Early in my career, people would ask me questions about my colleagues: Are they safe hands? Can you trust them not to make mistakes when hundreds of millions are moving? Automation makes the counting easy. But it does nothing to answer that question. In fact, it makes it harder, because the more we hand off, the easier it is to stop paying attention.
So here's my bet: After automation, the scarce currency in finance won't be dollars, yen, or pounds. It will be trust—built slowly by getting it right, and lost in a single wrong number. When execution is free, the people worth the most will be the ones with safe hands: the ones who still, personally, stand behind the number.
I have a stake in believing this, of course. But I'd bet on it even if I didn't—the world has trusted its scribes for 5,000 years, and it still does.
