Keep pulling the thread on Nufar Gaspar.
The quality of a leader's personal AI usage is the single biggest predictor of how effectively their teams will adopt AI.
Organizations where the CEO is the most proficient AI user tend to be the most advanced AI companies.
Leaders who delegate AI strategy without personal usage often either underestimate the technology's capabilities or set unrealistic expectations for their teams.
Using voice input tools like Whisper Flow for interacting with AI is superior to typing because it allows for more unstructured, intuitive thinking to be captured.
The latest advanced models, such as Anthropic's Opus models and recent GPT versions, are highly effective at processing unstructured and nonlinear spoken input.
A recommended technique for AI-powered research is the "wisdom of the crowd" method, which involves sending the same query to multiple models, aggregating the points of agreement, and investigating the divergences.
AI models are more effective at verifying information and fact-checking an aggregated result than they are at generating factually correct research from scratch.
For strategic decision-making, users can create a virtual "board of advisors" by instructing an AI to adopt different personas and debate an issue among themselves before presenting a conclusion.
To make an AI write in a user's specific voice, a "style profiling" technique can be used where the user provides their best writing samples for the AI to analyze and replicate.
Modern AI tools with built-in connectors have made it much easier and more reliable to gather and synthesize data across multiple, disparate company systems.
Before committing to building an automation, a user should first test the process manually and repeatedly for a period like one or two weeks to refine it.
After a user has successfully implemented individual AI assistants for specific tasks, the next logical step is to build a "chief of staff" AI that can orchestrate across all of them.