I consider myself to be a Generative AI (GenAI) power user. I spend about $60 a month on a handful of tools. I typically max out my top-tier responses on a daily basis. And it’s totally worth it. GenAI is one of the things that I credit the most for even thinking about becoming a SaaS (Software as a Service) entrepreneur. It literally changed the game in terms of skills I could access, and the speed at which I could prototype.
Six months ago, I wasn’t thinking about SaaS companies. And now I have three. Let’s get into it.
1. Why? On-demand fractional support
Over the past six months, I’ve started to make a transition from freelance contractor to solo (and sometimes partner) entrepreneurship. A lot looks the same. Like, I still spend a majority of my time working for clients for whom I provide a specific set of knowledge and skills. Like instructional design. Or education content development. Or dashboard building.
And…there is a lot that’s different.
When I put on my SaaS entrepreneurship hat, I’m suddenly no longer just playing a single role. I’m a software developer, data engineer, marketer, graphic designer.
I’m quickly out of my depth in many of these roles. But, I’m not going to hire anyone - I’m bootstrapping my companies. This is when GenAI comes in: on-demand fractional support.
So, what is on-demand fractional support? Well, Google doesn’t provide me with much in terms of a definition, so let’s go to the stars of this edition:
ChatGPT 4: “The term "on-demand fractional support" refers to a business model where specialized services are provided on an as-needed basis, allowing clients to access professional expertise or resources without committing to a full-time position or long-term contract. This approach is often utilized for accessing senior-level skills or specialized knowledge in areas such as marketing, finance, human resources, or IT.”
Claude 3 Opus: “"On-demand fractional support" refers to a business model where companies or individuals can access specialized expertise or services on an as-needed basis, without having to hire full-time employees. The term "fractional" implies that the support is provided in smaller increments or fractions of time, rather than a traditional 40-hour work week.”
For me, GenAI provides on-demand (like instantaneous) access to most of the kinds of fractional support that I need as an entrepreneur (I don’t use it for legal or finance support). This might range from front-end development to a social media marketing campaign to SEO strategy.
Now, talking to an expert for each of these needs would make a lot of sense. But I often need that information immediately, and don’t really have enough questions to warrant the time and cost of finding and contracting an expert. At some point, it will likely make sense to contract with actual people in these roles (or, maybe even hire someone full-time). But for now, I get the on-demand fractional support I need from GenAI.
2. When? Once it makes sense
I never start a project with a GenAI tool. I’m sure you could. But I don’t.
I start with a notebook. And a pen. I like the feeling of writing without the distraction of a million apps. There is something supremely satisfying about ink on paper. It’s as if the ideas are pulled by the pen into the pages.
This experience typically lasts for several minutes - whether it’s thinking about the potential use cases for my new GenAI coaching tool, or how I want the data to flow for my dashboarding app. I almost always start with the headline: “Brainstorming.” And I almost always end with a question or an idea that I need to explore more. This is when it makes sense for me to open up a GenAI tool.
It’s usually around something that I don’t quite have a handle on (like social media marketing) or an idea that I want a second set of eyes on (like, pairing a Flask backend to a React frontend). I’ll get a bunch of new ideas - sometimes it’s things I’ve thought of - but almost always there is at least one idea that pushes my thinking. And I like that, especially as someone who mostly works alone.
For strategy support, my prompts typically move from high-level questions to more specific wonderings. I might hop on Google to look something up or upload a resource that I’ve already been using. I build more and more context until I feel like I have a decent enough grasp to try something. Then I action plan. Or, actually, I ask for an action plan based on my line of questioning. The conversations are typically pretty quick - I come in with a question and leave with a plan.
But things look pretty different when I’m working on something more technical - like programming. These conversations are often much longer, include a lot of uploads or back-and-forth. There is usually a need to explore other ideas in places like StackOverflow or specific documentation. And, because many GenAI tools are not up-to-date, there is a lot of testing and iteration. These are moments when GenAI feels like a helpful, more senior colleague - except that I can ask as many questions as I want and they never try to leave.
I should also mention that GenAI plays an important role in the actual products that I’m building. Two of my three applications (and potentially the third) leverage GenAI under the hood via Application Program Interfaces (APIs). These allow me to automate the power of GenAI for specific use cases. More on this to come in future newsletters, but if you want to check out an early version of one of my tools, head to zentor.ai (contact me for a free login).
GenAI can do a ton of other stuff - like summarize, synthesize. All that stuff. And I use it for all those things, too. But for me, as an entrepreneur, GenAI is my ridiculously knowledgeable co-worker that jumps in with exactly what I need when I need it.
3. Which tools? It’s a bunch
When it comes tooling, there are several options - and more on the way. I’ve been using a couple consistently for the last few months, and am starting to dabble with a couple more. Similar to my approach for “when”, the “which” depends less on the tool and more on my purpose.
General Purpose Workhorses
ChatGPT from OpenAI was my first introduction into GenAI tools. I remember my first experience. I created an account, asked a question, and was blown away. Like, whoa! I probably only spent a few minutes using ChatGPT that day, but quickly started to integrate it into my workflows and signed up for a paid account within a week.
ChatGPT been my go-to for all kinds of conversations from strategy to technical support to under-the-hood development (I sometimes even ask it questions about raising my two boys).
Two things I love:
GPTs. In late 2023, OpenAI released the GPT feature - a way to build your own AI with options like just using information you provide. I like the GPTs for a couple reasons. First, I can silo off my content from the rest of the Internet. Just want it to synthesize a set of four articles? Done. Want to understand a set of standards without anything else? Done. And second, I often want to come back to the same conversations - with the same base of information - over and over again. GPTs allow me to do that without needing to constantly re-upload and re-design my prompts. Genius.
Inline Image Creation. What’s the phrase - an image is like a thousand words? I don’t know, but images are certainly helpful for quick prototypes. And having an inline image creation option where I can quickly pivot from a conversation to having an image generated is fantastic. I literally did that as I wrote the “when” section above.
And while ChatGPT has been awesome, I’ve recently been spending more time with Claude from Anthropic. Claude has a similar layout to ChatGPT, but lacks the ability to create images. I also tend to hit my daily number of responses faster. And it’s slower to initially process my question.
But here’s the thing, I actually like that slower response. Especially when I’m programming - which is my favorite thing to do with Claude. It gives me a second to ask my question, then refer back to my own code, then come back for the response. And I love the way it lays out the responses, including a clear explanation of the code:
I typically have both open when I’m working, and my sense is that I’ll continue to use them in parallel - at least until something else blows them away.
Specialized Tools
Now, if you’re a software developer, you might be saying: “Stop copying and pasting your code!” And yes, I totally get that. Which is why I’m starting to work with GitHub’s Copilot. It integrates into VS Code, my editor, and allows me to ask questions and fix bugs without leaving the program.
I love the questions part, especially when I’m trying out something new that I don’t totally understand. But the fixing the bugs part is a little nerve wracking. Because it often - at least in my experience - doesn’t fix the bug. For me, being able to read through the solution in a different tool before I make any changes in my own code, feels better. I feel like I’m making more informed changes. But I know it’s inefficient. So…more on that to come later.
My final tool (at least for the moment) is Midjourney. Similar to Copilot, Midjourney is a specialized AI tool - it just does images. And it does them really well. Like, whoa! For each prompt, you get four images and you can choose from them. When I first created the “multiple hats” image above, I got:
I wanted something similar to the first image, so I asked for a variation. It’s pretty quick and the images - whether they are abstract like the one above or a more photographic image like the dogs below, are amazing. I’m currently working on a targeted short-story application for parents of toddlers (like me), and I imagine Midjourney will be a part of that…journey.
Much like my learning, my use of GenAI is constantly developing. I’m sure I’ll be back with updates in the coming months. Thank you for reading and for sharing this with your entrepreneur friends - or folks who are curious about about GenAI.
Let’s go.
Love it! Have you ever used the voice chat feature in ChatGPT? I find that being able to have a conversation using my voice is a great way to practice conversations and get feedback in a way that is more natural than typing them out. I think you can only do it on the mobile app, though.