The problem with AI tools in the corporate world right now is that the biggest companies are overcautious about deploying the technology, fearing it might go off the rails. But the bold startups promising to take full advantage of the powerful new tool are selling products that actually do go off the rails.
Earlier this week, Writer CEO May Habib gave me a demo of her company’s new “Action Agent,” a tool that attempts to find a middle ground, giving employees at big companies the ability to use AI in a way that feels limitless but is also insulated with guardrails that ensure they can’t do any damage.
The software creates virtual computers on the fly and then gives AI full rein over those devices — something that would be extremely dangerous if it were operating on a corporate laptop. You could ask it to browse the web and take actions on your behalf, like filling out forms. And you can make those tasks repetitive. Something like: each morning, go to these seven different websites, gather certain information, put it into a chart, and send it as an email attachment to a single person.
Where this gets interesting is when employees start pushing the boundaries. Some employees could try to automate a lot more things than they should, or get around security restrictions to wield even more power. Then, companies will have to decide whether to fire those people or promote them.