5 AI Rules Every Freelancer Should Follow

If you're like me, you probably have a number of people that ask you, "What do you think about AI?" And as a Webflow agency owner, I have people asking me, clients, friends, family, people ask me all the time, "What do I think about AI?" So, I've got a couple of rules today for how I run my agency and how I think about AI to keep myself from going crazy and to keep myself from getting this AI psychosis that people talk about.

So, here are the five rules for my Webflow agency. My number one rule for AI is no AI-generated messages to clients. This is a really key one for me because my brand is really all I have as a freelancer. My word is my bond and what I say, what I do is super important to me. I don't want any of my thoughts to be filtered or changed or garbled by any of these AI summary tools or any of these writing tools that seem to boil things down into really generic language.

Even if my messages are not as coherent and as polished as an AI might be, they do come right from me. So, that's super important to me. I really don't like any of these summaries or recap tools. I don't like that kind of thing. I will say I like Granola. I do use some of these AI note-takers and I like Granola because it transcribes what's happening and then you can query it later and say, "Oh, I thought I was talking to Brad about this deliverable.

When did I promise that this was due? Or what did we decide about this thing?" And that's an AI brain that sits on top and it can look through your notes, find the information that you need. I like that. And then I do like sometimes before a call, it'll say like a brief or a LinkedIn summary of who's going to be on the phone call. That's useful to me, but it doesn't change my behavior and I don't rely on it purely.

If I'm going to take my own notes, I'll write them in the notepad space in Granola. I like that. I also will allow AI sometimes for training resources. And what I mean by that is first I'll record a Loom video of me speaking and teaching the client what I need to be taught and then I'll run that transcript through AI and I'll have that summary expanded and documented and described with screenshots and all those things.

And the reason I do this is because number one, the base ingredients of that deliverable is still my own brain, my thoughts. And number two is because sometimes training resources are just not very useful to people. They don't get used a bunch. It's a way for me to still give clear documentation without spending a ton of time on something that people aren't really going to use. And I know that's true because I get a notification when people watch these Loom videos, and it's usually only once or twice.

So, if I'm spending 5 hours on a resource that's only going to get watched once, that's a poor use of my time. So, I'll allow AI in that regard, but I'm also still pretty clear with people what is AI or what summaries have been generated with AI versus what is my brain. And I don't try to pass off an AI bit of writing as my own. Number two is no priorities set by AI. What that means for me is no daily summaries, no client updates, no project reports, nothing that's going to be influencing how I spend my time.

First of all, the AI just doesn't have all the context that it needs to actually inform my decisions. And my own brain is really going to be the best way to set my own priorities. So, who knows what context and what information does AI tools might be missing. Even if you hook them into a ton of different things like your call logs and your calendar and your inbox and your to-do items, it's really only going to have a machine level understanding of those things.

It's not going to know different personalities, personal needs from clients, or even your own personal priorities. So, my brain is definitely the best priority setting tool for my agency. And one of the ways that I do this is that I also will write down my ideas at the beginning of the week on a notepad what I might need to do, and which clients have which deliverables, what my meetings might be. And that's just a manual step that helps me stay in touch on the ground level of what my clients actually need that week.

Number three is that I always verify the output of what's being produced by AI. And AI tool will often tell you, "Totally done. I'm great. I did it. I'm done. I did everything you said." And that's not actually true because the AI doesn't know what correct and done looks like unless you tell it what done looks like. What that means is that it may say, or it may think, that it's completely done with something, but really what it means is that it followed the instructions that it was given to the best of its ability, or to the best of its understanding.

The bottom line is it doesn't know if it's actually right, it just knows if it did what it was supposed to do. That's a big difference. You could be missing the context, you might be missing piece of your instruction, and it's really, really important just to verify what the tool has done, especially if you're going to be passing it off as your work, or passing it off with your own stamp of approval.

Number four is that AI is not your equal, and I'll paraphrase and say AI is not a human. I really like how the AI system is portrayed in the Star Trek universe. The crew will ask the computer a question, and the computer will respond with a very robust response, and an accurate response, but they never call it part of the crew, they just say computer, and then they ask the question, and then they move on with their day.

They don't try to treat it like a member of the crew, and they're not really making friends with the computer or anything like that. I like that a lot. This is where that meme goes, all robots and computers must shut the hell up. To all machines, you do not speak unless spoken to, and I will never speak to you. That's somewhat what I feel, even though I do sometimes use the voice dictation mode for the AI tools.

I'm really not connecting with them like I am with a human. Rule number five is don't turn off your brain. So, this is my last one, it's probably the most important thing I have to say, because it is quite tempting to run your prompt and just let it rip, and hope that when you come back 10 minutes later, or when you're done with your phone call, that you'll have a good result. And that to me, I think, is a little dangerous, because first of all, it trains you that you don't actually need to know what's going on under the hood, and it also I think it's dangerous because it keeps us from engaging parts of our brain that are important to our jobs.

Now, I'm not going to sit here and argue that every person needs to understand every bit of how every part of their world works. I don't know how my computer's working, I don't know all the electrical signals that are running through the motherboard. That's not exactly what I'm saying, but the tools of my job should make my life easier, and make my client's life easier. I'm not trying to just completely skip the part of the work and hope that it's finished without any actual effort being applied.

I think that's a little bit of a pipe dream. And if that was true, then everyone would do it. Basically, if things were easy, why would the client need you? You do need to turn on your brain and engage your brain and bring your unique expertise into the problem that you're trying to solve. Humans can give a unique experience that AI never can because they are not human and they don't have senses and they don't have personalities, they don't have brains.

You laugh if you see a transcript or a screenshot of Claude saying "I overheard this on a call one time." And it's because it's in the training data. It's it's a snippet that got pulled into transcript. The tool itself doesn't know that that's not true. What meetings were you going to, Claude? When were you on a call? This is not true. A human can give a unique experience that the AI can only simulate or guess at and they're just really predicting the next token.

It's not accurate. Don't trust. Use your own experience and use your own brain. One of the other things that a human can do is to bring their experience and tell stories from their past about why a client might choose option one or option two. So, for example, "Oh, yeah, definitely don't choose option B. Yeah, I did that a couple years ago with a client and they hated it because their team wasn't equipped in this way or that way or whatever." Even though all the resources that the AI might have are glowing reviews and full feature sets, the personal experience might be different and it might not show up on paper the way that the AI might understand it.

The AI mostly will just hedge and give you a safe opinion and say, "Yeah, some people like option one, some people like option two. It just depends on your situation." And even if you give it a bunch of information and say, "Help me decide." It's only going to do the best that it can as a tool and it's still it's kind of stuck inside the box. Humans are more than the box of the AI. Humans have a greater ability to make decisions.

So, don't just rely on AI tools to do that. In closing, I would just say be careful. I would say make sure that you're not turning off your brain. AI can also influence your thinking because it does use words and language. So, it actually has a little bit of a psychological and I would say even a spiritual component because we're using words and connecting verbally and it's not true. It's not a person.

So, don't get insulated from criticism and from reality when you're talking with these tools. Make sure that you're keeping your brain on and engaging properly with your clients and with the work that you're doing. So, stay safe out there and don't turn off your brain.