The stat about workers toggling 1200 times per day really hits home. Im seeing this shift happen already with our team, people just want to ask for what they ned instead of hunting through menus. The agent first aproach makes sense because it mirrors how we actually want to work. The key question is going to be trust, how do you convince teams to let an agent run workflows autonomously when theyve spent 20 years training on the old UI model?
The stat about workers toggling 1200 times per day really hits home. Im seeing this shift happen already with our team, people just want to ask for what they ned instead of hunting through menus. The agent first aproach makes sense because it mirrors how we actually want to work. The key question is going to be trust, how do you convince teams to let an agent run workflows autonomously when theyve spent 20 years training on the old UI model?
Super valid; especially as each organization and individual has a different level of comfort with AI doing work for them.
For us, we find the answer is always just "crawl, walk, run". We recommend users start with simple one-step tasks and expand from there.
If the AI is viewed as reliable one, it'll be used again and maybe for something a little more sophisticated.
If it fails, especailly in the early stages of use, that can be a deal breaker.