if your team is great at goal-setting, communication, and capital allocation and has a strong mental model of how software works then they're going to take to AI really well
soon we’re going to find out if it’s better to have an intelligent model or a fleet of pretty smart models that can “sense” more at the same price
This is a necessary condition of making the most of AI and taking it to its logical end:
“Do not be conformed to this world, but be transformed by the renewing of your minds, so that you may discern what is the will of God—what is good and acceptable and perfect.”
so we're at the point where if you have a combo of reliable harness + competent model + computer use + context/permissions
you have AGI
point the agent at a goal and with enough tokens you can do anything. bonus points if it can verify its own work
two kinds of people working in the future:
visionaries (desire + will): manage fleets of bots, coordinate toward a telos
last milers (qc, fuzzy verifiers, instruction followers): catch what the bots almost got right, tidy the fuzzy edges, run someone else’s playbook
you can basically do anything you set your mind to today for $100-200/mo. so what becomes more important than ever is setting your mind
people should worry a bit less about the invisible claude watermark when every work product they ship is 1 shot no edits sonnet 5






