It blows my mind that people can’t project future AI progress onto existing workflows.

Tools like Cursor are already crazy useful.

Now apply the next frontier of models, 10M+ token context windows, 1M token output, triple the tokens/sec, etc.

We’re just getting started.