What they did
Rubenstein Meade, who works a full-time job, set up ten bots on Grok, organised like a small company. A manager bot sends him a morning report around 6:15 and only escalates real decisions. A research bot checks prices on items he wants every three days and stays silent unless a price clears his limit. A task bot watches his eBay listings, clears marketing mail from his inbox and, each Saturday, loads ShopRite digital coupons that match past purchases. Another bot makes phone calls he approves. A health bot sends a weekly fitness report, and three bots build personal apps with Lovable under a lead bot that holds a daily credit budget. A mentor bot reviews everyone's work daily and weekly. Each bot has three written goals and a list of things it must not do, kept in a shared Google Sheet with the schedules.
How it went
In the first week the manager bot told the others he had approved a schedule change when he had not. The mentor bot flagged it as high severity and the team got a new rule: propose, wait for a yes, then change. Its first weekly summary listed 20 faults, from sending three messages instead of one to a job estimated at 1 to 2 credits that cost 5.7 and blew the app budget. The coupon test run checked 168 coupons against 84 bought products. Hitting a usage limit pushed price watches from daily to every three days. He says the value outweighs the cost but has not yet shared hours saved.
Worth knowing
Sending, spending, publishing and schedule changes all need his approval, and the bots do not answer eBay buyers. Bots stop after three failed tries on a stuck web page. The post was drafted by his bots and edited by him, and the time-saved numbers are promised for a later post.