@tjrobertson52

I think we're about to see another step change in the way knowledge work is done, similar to the step change we saw when ChatGPT was released. I wanna talk about what this shift is, why I think it's coming by the end of 2026, and what you can be doing now to prepare for it.

Large language models like ChatGPT have already disrupted a number of industries. But for the last two years, people have been saying that AI is gonna take all the jobs, and it just doesn't seem like it's happening. While AI has made a lot of jobs more efficient, people are still hiring.

I think this has left a lot of people thinking that AI is just a bunch of hype. The expectation that a lot of people have is that if AI was going to take a large percentage of human jobs, then surely by now it would have taken more of them, and that seems reasonable. But that's not quite how this technology works.

It doesn't slowly and consistently replace workers over time happens all at once.

5 wasn't that much better than GPT 3. So why was ChatGPT such a big deal? It's because it passed that good enough threshold.

GPT 3 just wasn't quite good enough to be useful. 5, it became useful and immediately displaced a lot of jobs.

Now since then, we've seen a handful of similar shifts into just useful. We've seen it with image generation, we've seen it with presentation generation and an AI voice generation, but none of these had as big of an impact as ChatGPT.

The step change that I'm predicting by the end of 2026 is what we're currently calling agentic browsers or computer use. The current versions of this are like perplexities comet browser, ChatGPT agent, or Claude computer use. Right now they're mostly just fun toys.

There aren't many practical applications for them. Currently, you might say these tools are in their GPT2 era, and I think by the end of 2025, they'll probably reach GPT3 status, but they will still be effectively useless until they have their ChatGPT moment.

Eventually these agents will be able to do essentially anything a human can do on a computer. However, they'll probably still have some of the same limitations that large language models have today. Mainly, they'll have trouble seeing the bigger picture.

So they're not gonna replace your senior level employees anytime soon, but I do think they'll be able to do any task that you could currently hand over to a new hire with a well written SOP. And that's why I think the best use of your time right now as a business owner is to write really good SOPs for all of your processes.

And the way we write SOPs for large language models will probably be a little different than how we write them for humans. But taking a well written SOP for a human and reformatting it for a large language model is the kind of thing that large language models are really good at. Don't wait until we know how to write SOPs for large language models.

Just start making really good SOPs for humans now. Document every single process in your business that can be done on a computer and you'll be perfectly positioned to take advantage of the next ChatGPT moment.