Summary
Nick Saraev shares seven habits he says separate power users of AI from everyone else, arguing the gap comes from how people talk to models rather than which model they pick.
The first tips focus on speed and capture. He recommends dictating instead of typing, since people speak at roughly 150 to 200 words per minute versus 30 to 50 typed, and mentions free dictation tools and an app called Hex. He also describes an always-on voice stream during meetings, powered by real-time voice models, that transcribes conversations and kicks off real tasks in tools like Notion, Asana or Monday while the meeting is still happening. He estimates such a setup could cost a dollar or two a day.
The remaining tips cover how to get better output. He suggests asking AI to build visualizations with Three.js, HTML, Blender or Unreal Engine rather than relying on text, having the model interview you to surface missing context, and asking for flaws instead of opinions to work around the tendency of RLHF-trained models to agree with users. Finally, he advises testing whether AI can do a task by simply asking it to try. The video suits anyone looking for practical ways to get more from frontier models.
📺 Source: Nick Saraev · Published October 14, 2026
🏷️ Format: Tutorial Demo
