Adam Sand breaks down why most people's AI outputs fall flat: they skip steps. He maps a five-phase workflow from ideation through distribution, arguing that process matters more than which tool you pick.
If you've ever typed write me a blog into ChatGPT and got disappointed, it's not because it's bad, but because you skipped 80% of what actually makes it work. I'm Adam San, CEO of the Roofing business partner. In this video, we'll break down the full 5 phase AI process. Why the tool doesn't matter, but your approach does, and how you can use AI in ways 99% of people completely overlook.
Where do people go wrong with AI? They don't know the five phases this they're used to. Using AI to write something or to answer a question. In many ways, having a chat with ChatGPT, that's like 90% of the society's interaction with what AI is. So it's where you go into ChatGPT or Claude or Gemini or some other AI tool and you say, write me a blog or generate me an image. You don't get the desired outcomes because you're missing four other steps. But to really get AI to work for you, you have to go through five phases. You have ideation, research, composition, that's the one that you're doing right now, usually refinement, and then distribution. The tool does not matter as much as the process, and the process does not matter as much as the adoption. And the adoption is those five things. The reason that you want to do it this way is because you can be using AI across all five of these steps and you will get better outcomes this way. The first one is ideation. You're coming up with an idea, how to ideate, how to think about, think of what your idea might be to kind of really get clear on what that idea is. Then you're going to do research, research around that idea. So you then say, I want to figure out how to do something like that. When you do your research, then you can start to put together how you're actually going to approach the composition step a lot better. Then you get into composition. This is the part that everybody is very familiar with. ChatGPT. Write me a this, you know, edit my this, look at my this and edit that. Then there's the refining and editing portion. It's so easy to get AI to create a mass amount of information. Well, how do we trim that down? And then distribution. And AI can help with distribution now more than ever when talking to AI, whether you're doing in chat or if you're just talking to AI, like I want to kind of say you can assign artificial intelligence a role, you can give it a personality, you can give it a job to do. You can give it a style, you can assign it a profession, you can give it a perspective and then you can also ask it to format the output. OpenAI kind of was the one that gave all of humanity kind of the understanding what AI could be. But it's actually been around since like the 50s started just by identifying pictures of cats and dogs. Then you get this like massive explosion about three years ago now when OpenAI went from GPT to GPT2, then GPT3. And GPT3 is where they added chat to it, made it so you could talk to it. And that was like the big breakthrough that made AI kind of accessible to everybody. And by accessible it was free and people were blown away. Then they charged $20 a month. So you get access to the smarter models or larger context windows kind of multimodal things. You could send it pictures or you could get it to output graphs or you can make it generate videos.
