Welcome to the 300 or so new subscribers, we’re now a community of 12,000, wild!
The world of AI is moving fast –too fast to keep up. Every day there are new AI tools being released, from tools that turn your audio ramblings into refined blogs posts, and tools that scrape your website and create blog posts, to tools like AutoGPT which can (supposedly) run a business on autopilot. Disclosure: I tried to get AutoGPT to start an e-commerce business for me and it just got stuck in an infinite loop when I give it API access to my blog.
My default mode of thinking now whenever I start a new task is ‘How could I use AI to get this done better, and faster’. Writing this newsletter - can I use AI for that? Designing a new product, can I use AI for that, strategising a marketing plan - can I use AI for that? With so much noise, it’s easy to get caught in decision paralysis.
In this article, I want to break down how I use AI on a daily basis. I want to share not only where it helps me, but where it can be a hindrance. Hopefully, after reading this you’ll be able to learn from my mistakes and not waste tens of hours exploring dozens of different AI tools and workflows.
Writing
Not all writing is created equal. In my work at Lithium, sometimes I have to write in-depth articles, sometimes tweet threads, sometimes internal documents for my team. When I first started using AI, I’d take the same approach for every task, just bang a brief into ChatGPT and see what came out, after wasting a tonne of time on this, I now take quite a different approach…
Newsletters and long-form articles.
At the moment, I don’t like using AI tools for long-form articles or for my newsletter. The reason is quite simple, for me, writing is thinking. The act of writing is what helps me form ideas or clarify my understanding. If I’m just giving prompts to GPT on ‘write an article about stuck culture’ or ‘break down this persons success’ I’m getting the output, but I’m missing the understanding.
It’s the difference between following a recipe step by step or building the confidence to cook from fundamentals. Sure, recipes can help you build an understanding, but you’re never going to be a good cook if you just blindly follow them. The other reason I don’t like writing long-form stuff with AI is it just lacks personality. There might be a time when AI can scrape every article I’ve ever written and match my tone, but it’s not there yet.
Video Scripts
AI helps me massively with writing scripts for videos. I’m currently creating videos for a new YouTube channel for Lithium, focused on sharing insights in how to build communities. My process is quite simple,
Pick some influencer who has done an incredible job building a community
Get AI to flesh out an article plan
Add some colour, with specific examples from their Twitter or website (I haven’t yet found a tool that can do this well).
Use these bullet points as the basis for the video when I hit record.
The beauty of using AI for video scripts is that no one is going to read these scripts, and there is no downside in them lacking personality, that will get added in the recording process. On the flip side, they help give a video structure, just talking to a camera without any prompts can often lead to a mad ramble (sorry to any of you that have been exposed to that on my YouTube channel).
Twitter Threads
Last week I managed to make a connection with Greg Isenberg. Greg has over 300k followers and is an OG in the community-building space. Last week’s newsletter was a deep dive into how he’s built an empire off community-led growth.
I got Greg's attention through a well-made video and a catch Twitter post.
The output of this was Greg recorded me a 5+minute video going through questions I had to ask him about what we are building at Lithium. This is not only insanely helpful feedback but also opens up a bridge for further collaboration.
The kicker? ChatGPT didn’t help with this Twitter thread. I think many people think because Twitter threads are shorter and lower effort than articles, they are ripe to be displaced by ChatGPT. It’s exactly the opposite. Because there is so much noise on Twitter and because you only have a short amount of time to hook the reader, it needs to be authentic, insightful and unique. Sure you can use GPT to get a basis, but if you want to have an impact on Twitter, you need to do the final draft yourself.
Thinking
Idea-generation
For me, ChatGPT comes into its own when I’m thinking. My zone of genius is in the idea-generation phase of a project, but I can get a bit overexcited and bullish on any one particular idea. AI is my perfectly logical, right-brained teammate. She will challenge my ideas, show me gaps in my thinking, and help me form more robust concepts.
Below is the prompt I’ve used. Unfortunately, as this is for a top-secret product I’ll be offering you guys next month I can’t reveal the details!
Make sure to get yourself a Selma!
Refining
Perhaps where AI shines brigher is when you get into the phase of refining your thinking. A lot of my time is spent creating Product Requirements for developers. For anyone whose ever had to do this kind of work before, they are a fiddly, often soul-destroying part of the job of a Product Manager. You basically have to write out, in detail, exactly how the product should behave. I have a shitty attention to detail and I hate writing ‘em. That is until ChatGPT came along, now doing PR’s are one of my favourite jobs.
Here’s the output of ChatGPT when I asked it to act as a developer who had just received my product requirements and wanted to get some clarity. This might not mean much to you, but hopefully, you can see the general trend, it’s identifying edge cases and gaps in my thinking that I wasn’t aware of, I can then just answer these questions and the AI will update my requirements for me, with this new information, wild.
Of course, this is just one example of how it relates to a specific job role, the broader point I’m trying to make here is that whatever work you do, there is usually a point where you have to really dig into the details, GPT-4 is exceptional at doing that.
Design/Art
Last week I launched a community - AIFounders for solopreneurs and small business owners who want to use AI to get more done in less time. By the way, if you haven’t joined, it’s free and you can get access here.
I wanted to create a very simple landing page for my community that would let people input their email to get access. After buying the domain on Namecheap, I decided to use Framer to set up the site, it’s great no-code tool for site building. I knew I wanted a hero image to go alongside the text, but didn’t fancy paying for a designer to mock something up. Luckily, we now have Midjourney.
With the Midjourney product, I was able to create the below landing page hero image. Also while writing this article I realised the logo was a bit shit and re-did that on Midjourney as well! All in all this whole site, logo and hero image took me about an hour, and more importantly, it was fun. Getting a really cool image from playing with your prompts is actually quite rewarding.
Hopefully, that’s given you some insights into how to think about where to slot AI tools into your workflow. I have to admit that it is becoming a full-time job just keeping up with the advances in this field. Unless it’s something that really interests you just for its own sake, I’d recommend staying off Twitter and just letting the good stuff find its way to you through word of mouth or email, otherwise, you’ll spend half a workday on non-value-add process stuff.
Also Substack has a comments feature, so if you view this post online, I’d love to hear how you are using AI, or come say hi in our AI founders community!
WAGMI,
Tom