Why I stopped listening to AI gurus (Naval was right)
Naval just invalidated all the ‘advice’ that AI gurus sell to make you feel completely overwhelmed.
All the noise about having to use a new tool, or you’ll get left behind. When in fact, none of it really matters.
After watching the latest @navalpodcast episode, I gained more clarity on how I should use AI in my work, and here’s what I learned:
The tools, agents, or models do not matter
While everyone online insisted that you need to use OpenClaw, Hermes, Ralph Loops (and now /goal), Naval ignored all of it.
And he was right:
The days I spent troubleshooting all of my agents were completely wasted because I no longer use them.
I’ve tried plenty of prompts from AI influencers, but they didn’t give the output that I wanted.
If you want to use AI in a way that works for you, it doesn’t matter what tools or models you use, because all of them are interchangeable.
You will waste so much time getting it to work, and the opportunity cost is producing more meaningful outputs.
Or creating more value for your employer or your own business.
These are huge time sinks, but everyone wants to play with them because it’s the hottest thing right now.
Yes, it’s cool that agents can do all of the tasks you hate for you.
But so many things can (and will) go wrong.
You will get mediocre outputs or even worse, do something harmful that can’t be reversed (and the agent can only apologise).
The tools, models, and agents are all major distractions.
Most models are good enough to execute the tasks for you, so long as you give them this:
Focus on shaping the problem
If you want to become AI-native, the very first step is having a complete understanding of the problem you want AI to solve for you.
I call this the OWB of the problem:
Outcome: What do you want to achieve?
Workflow: What are the steps required to achieve that outcome?
Bottleneck: What blockers are present in the workflow that prevent you from achieving that outcome?
You have to be clear about the outcome, so the AI stays aligned with you.
You have to be clear about the workflow, so you can give AI clear instructions and it executes the task just like how you’d have done it.
You have to be clear about the bottlenecks, which are potential steps that AI can automate the task for you (especially if they’re soul-sucking and mindless).
This is the boring part about learning how to use AI:
You can’t automate what you don’t know. You have to become the subject matter expert of the workflow so you know how to direct AI to replicate it for you.
Only then can you automate the workflow in an effective way.
So if you want a clear path to becoming AI-native without all the noise, sign up for The AI-Native Sprint waitlist.
This is a 90-minute live session where I’ll help you install an AI system that works for your specific workflows.