Ohad Tevet and I have known each other since we were kids, and he was also the first person to buy my course. Today he is an employment lawyer with his own office, and what brought him to the podcast is the way he built that office, twice. Once as a regular office, and a second time as a system, after four months of reserve duty brought the first one down.
An entrepreneur before he was a lawyer
Before law, Ohad was an entrepreneur. He had a printed clothing business, which he sold when he felt he had taken it as far as he wanted. Today he only looks for lean businesses, with no logistics, no inventory and no thin margins. He ran a bar mitzvah preparation program for secular families, and later became a partner in a company that set up outdoor events. The story of that company is one of the best moments in the episode: the owner was about to close it and leave for his big trip, and Ohad offered him a cut of the profits while he ran everything. In the good months they issued dozens of payslips. When Ohad got his license, he sold his share.
84 payslips in Excel
The pain that pulled him into systems started during his law internship. A client who worked somewhere for seven years arrives with 84 payslips and attendance reports, hundreds of documents, and someone has to copy all of it into Excel and cross-check it. Ohad calls it ant work, and explains why it cannot be rushed: the human eye is simply not built to find the one small figure that does not add up. His first attempt to build a system for it got stuck on extracting text from the documents. Years later he joined a system built by another lawyer as the marketing and sales partner, and without spending a shekel on ads, using only webinars, it reached dozens of users.
Anything done three times becomes a system
When he discovered Base44, Ohad started building for himself. His rule is simple, and I think it fits every business owner: any task that repeats at least three times becomes a system. That is how he built an automation that reads the emails from Net HaMishpat, the Israeli courts system, and pulls out decisions and deadlines, because there is no open API and no lawyer should have to open every case file every morning. That led us to a point that matters a lot to me. When anyone can copy a product with AI, the real advantage is no longer in the code. It is in places that are hard to get into, like a regulatory approval that only you have.
Four months of reserve duty
Then came reserve duty. Four months away, and Ohad came back to an office that had fallen apart. The salesperson left, the intern left, and he had to start from scratch. This time he did it with Claude Code, building skills for the legal work and for marketing. By Claude Code's own count, in one month they did 250 net hours of work together. Ohad says that in his view the tool does better work than most lawyers, and in the same breath stresses that everything goes through him, because the professional responsibility stays his.
Automation versus willpower
We also talked about how much you should stay hands-off. I brought an example from my own life: a standing order that goes out every month without anyone asking me whether this month is a good one. A ten-second task is never really ten seconds, because it depends on you, and anything that depends on willpower breaks on the first day you are sick, busy or just tired. On the other hand, there are things where your involvement creates order, and those are worth keeping manual. We argued a little too. Ohad thinks most lawyers need a closed system that does exactly what they need, rather than opening Claude Code themselves. When it comes to professionals like lawyers, I agree with him.
Who this episode is for
For professionals who want to understand what AI can do for their practice, for anyone torn between building it themselves and getting a ready-made system, and for anyone who knows the moment when a business depends on people who are suddenly not there.