There's a quiet revolution happening inside Block, and it's not wearing a cape or shouting from a keynote stage. Managerbot, the company's new proactive AI agent for Square sellers, is the first real product to justify the kind of existential bet Jack Dorsey has placed on artificial intelligence, a bet that cost 4,000 jobs and reshaped the company's entire identity. But for the small business owner running a café or a retail shop, the stakes are far more personal than corporate strategy. This is about whether an AI can genuinely make their week easier, or whether it's just another layer of software demanding attention and trust.
What stands out about Managerbot is not that it answers questions, plenty of tools do that. It's that it watches, waits, and then acts on the seller's behalf, flagging an inventory shortage before it becomes a lost sale, or drafting a win-back campaign for customers who haven't visited in a month. That's a meaningful leap from the reactive chatbot model, where the burden is always on the user to know what to ask. For a seller juggling payroll, schedules, and supplier calls, the difference between "ask me anything" and "I've got this covered" is the difference between another chore and a genuine relief. The fact that every proposed change requires explicit approval is the right call, too, not because sellers can't handle autonomy, but because trust is earned in small confirmations, not in silent automation.
Still, we'd be naive to ignore what this product is really doing beneath the surface. Every seller who adopts Managerbot is feeding more of their business into Square's ecosystem, moving payroll, time cards, and scheduling onto the platform because the agent works better with more data. That's not an accident, and it's not a bug. It's the strategy. Avé is transparent about the compounding value, and we don't fault him for it. But the same tool that protects a seller from running out of stock is also, deliberately, the thing that makes Square harder to leave. That's a trade-off worth naming, because small businesses have been burned before by platforms that start as helpers and end as toll booths.
The real test for Managerbot isn't whether it can forecast weather-driven demand or optimize a shift schedule, those are solvable problems. It's whether it can avoid the pitfalls that have tripped up Block's earlier AI experiments, from the $80 million fine tied to Cash App compliance failures to chatbots that told customers to close their accounts. Avé insists the harness layer, the engineering around the models, is where the magic lives, and he's probably right. But no amount of prompt tuning erases the risk of a probabilistic system making a confident mistake in a regulated domain. The guardrails are there, and the human-in-the-loop design is a step in the right direction. What we'll be watching is whether sellers actually feel protected, or just managed. If Managerbot delivers on its promise, it won't just be a feature; it'll be the reason Square becomes the operating system for Main Street. That's a future worth building toward, but only if the trust holds up under the weight of the company's own ambitions.