Permissionless Prompting
Jack Butcher popularized the concept of The Permissionless Apprentice. AI makes this 10x easier and 10x more powerful for those willing to prompt (and share the results).
Many of you are old enough to remember the popular website “Let Me Fucking Google That For You” (LMFGTFY). Passive-aggressive tone aside, it was useful and necessary.
Because I’m a “technical” person, many friends and family would use me as their Geek Squad. What they didn’t know is that I just took their question 50% of the time, put it into Google, and summarized the result.
In fact, a joke among software developers (even Senior devs and engineers) is that they are just better at searching Stack Overflow to figure out how to fix bugs better.
So, using LMFGTFY was a reminder to the person asking that… the world’s knowledge base was also at their fingertips… they just had to use it.
Enter AI
Tools like ChatGPT and Perplexity are 10x better than Google Search. It can better understand what you’re asking about, skim 10-30 websites simultaneously, assimilate details across all of them, and summarize them to your liking.
It’s hard to stress how powerful this is.
Better yet, these services built LMFGTFY functionality right into the product.
Enter Sharing
I run Deep Research reports on OpenAI anywhere from 4-10 times a day before meetings with new clients, investors, etc. It took me 1 minute to ask my question, and then I returned 20 minutes later to get a 20-page report with 30 citations and a 1-page exec summary. I skim the summary, read on for more details, and viola… I’m now prepared for the next meeting.
But the killer feature is that I can share this with other colleagues.
I frequently share all the recon I just did via text or Slack, so they are equally informed and on the same page as me.
These reports are gold. And given they 10x my preparedness in 10x less time than it used to take Googling, I feel like I’m just starting to scratch the surface of what’s possible.
Enter Permissionless Prompting
Like LMFGTFY, I see colleagues asking some good questions all the time but never taking that back to AIs/LLMs to go deep.
So I just copy, paste, and share the results.
At least 50% of the time I get a “holy shit” response.
I asked the AI model to give summaries, different perspectives, pros and cons, section summaries, etc. All that information is right at your fingertips in minutes. It’s like having a dedicated research analyst for free.
I’m starting to get a little more aggressive with this. When people post things on social media, my go-to move is to pass along their questions and share the results. They didn’t ask me to do it, but I’m curious. And if I already found a wealth of knowledge that takes me 5 seconds to share back, it makes sense to do that.
I hope that, like the old LMFGTFY days, enough people get hooked by that first experience that they start going to the tools themselves.
Yesterday’s Example
A neighbor of mine had an AC issue, and they had a unit with Freon that is no longer sold in the US. Rather than face an expensive $10K replacement, they were asking what they could do instead.
Using Deep Research, we found a mixed set of options. A limited number of suppliers were certified to produce/sell. Still, there were also reports of scams that resulted in damaged units. Overall, AI was able to scan forum comments across multiple websites and give 2-3 viable options that appeared to have legit 4-star reviews (5-star reviews can often be paid reviews, so they are less trustworthy).
Usually, the cost of doing this digging would be too high to bother. But I could submit this when the phone call started, and I had it in hand before we were done chatting.
Is this a silver bullet? No. AI is still imperfect, and some people don’t want advice, they just want to be heard. But for those looking for help, this provides a swift, accurate, and cheap way to solve problems quickly.
And after all… you don’t need anyone’s permission to prompt to your hearts content. Sharing is always optional.