Oh, it’s AI all the way down — the rise of the 1-person $$$m business.
Update: the second law of thermodynamics has had its way with this. Given how rapidly this space has been evolving, I have not been surprised that my code has increasingly become obsolete, and I am not in a position to prioritise its survival. However, since writing this article, OpenAI announced the launch of “GPTs”. I have now been able to build evAIuate as a custom version of ChatGPT, which can be found here, and was built with about 5 minutes of effort. Mind blown…
evAIuate.com now links to this GPT since the launch of OpenAI’s GPT store. Enjoy!
“Can you imagine a world where a single individual could run a multi-million-dollar business from their living room? Welcome to the era where AI is the secret weapon.”
The entrepreneurial landscape is evolving. Gone are the days when establishing a business meant having a big team and significant capital. Solopreneurship has been on the rise, with one-third of UK adults thinking of starting a business; and you can see why — the flexibility it could potentially offer, the world we live in becoming increasingly digital and platforms like Upwork have made it easier than ever to take the leap. And that is even without thinking about our favourite term of 2023, Artificial Intelligence (AI); however, it’s with AI that things get really interesting.
As a small firm with big ambitions, you can imagine that we have been excited to see how we can use AI in our own day-to-day activities. And, by understanding how AI can help us, it can help us better support our portfolio companies as we gain a better understanding of how this foundational technology could change the business landscape. As such, we have been using it since the beginning of the year to help with a number of elements of our business, which can be split it into three key areas:
- Coding: It is in this space where I continue to be blown away. I am a hobbyist coder, having picked up a bit of Python over various lock-downs during the COVID-19 pandemic, and GPT-4 (and most recently Cursor) has given me enough co-pilot assistance to build internal tools to help streamline our processes. So far, we have optimised our deal-sourcing scripts that power our data insights, standardised deal screening reports and removed countless repeatable, thankless tasks that no human wants to do.
- Content Production: I’ll let you into a secret. ChatGPT helped me write this (although I’m quite proud of the title inspired by the late greats Sir Terry Pratchett and Carl Sagan). We don’t write enough marketing-style long-form content to pay for one of the more focused services like Jasper, but writing content with the right prompts can significantly increase your word count velocity and is also likely to be more creative than you think.
- Market Research: As an early-stage venture investor, having a deep understanding of the problems our founders are looking to solve, and therefore the markets in which they are building businesses in, is a key part of our investment decision-making process. Services like Perplexity and ChatGPT (with web browsing enabled) supplement and supercharge our ability to better understand an industry, complementing our desk research (collecting and analysing market insights and reports) as well as performing expert / reference calls to build our knowledge on the back of those who directly have the problem or deep domain expertise.
Having only previously written Python scripts, but now with my newfound coding superpowers, I had a fun question, “Can I build an AI-powered production-ready Web App?” Could someone who is technically inclined but completely unfamiliar with Web App development actually build one and launch it into the wild?
This was going to be a big test to see if AI could genuinely become my “technical co-founder.” Read on to hear of my joys, frustrations and learnings as we built and launched our product.
Introducing “evAIuate”
As an early-stage VC deeply embedded in the LSE entrepreneurial ecosystem, we see a lot of pitches, and unfortunately, we have to say no to the overwhelming majority of businesses that come to us as an investor. However, we want to be supportive as possible, and this led to the idea behind evAIuate. evAIuate is an AI-powered pitch deck analyser providing feedback to founders on their pitches and giving them recommendations on areas they could look to improve. To make it as engaging as possible, it actually rates your pitch out of 100 and scores the 10 most important areas out of 10, providing feedback on what was good and what could be improved. Now, in the hands of entrepreneurs, I have been energised to hear that it is helping and providing important feedback for those who are looking to improve their pitches.
If you end up using the product — reach out — feedback is always appreciated and helps us shape our thinking and the product itself.
The tool is actually quite flexible and can evaluate any document in the eyes of any reader — screenshots below were produced as I asked evAIuate to provide feedback on this blog post as a reader with an interest in venture capital and AI.
Behind the Scenes: Building “evAIuate”
Being technically savvy and with some experience coding in Python, the knowledge gap I was struggling with was how to turn my idea (fairly easy to build in Python into a full-blown Web App). This was going to involve cloud infrastructure and a steep learning curve into a domain steeped in jargon and complexity. Enter my infrastructure co-pilot ChatGPT. I cannot underestimate the role this played in turning my Python script into separate front-end and back-end code, as well as helping me to deploy it successfully on AWS (back-end) and Netlify (front-end). It took a bit of trial and error to work through some of the deployment failures and infrastructure nuances, but ChatGPT was able to navigate this layman through the challenge.
Infrastructure aside, it was also the co-pilot to the functional codebase of my App as well. As part of this project, I built features I have no previous experience developing, including sending transactional emails, sending internal Slack notifications, integrating Stripe to take micro-payments, logging client and server-side errors and setting up the application database. And there is, obviously, the all-important OpenAI API connection as well. As of today, the main application is ~900 lines of back-end code (Flask) and ~1,100 lines of front-end code (React).
With the App ready tested, working and ready for go-live, I was ready to see how AI could help on the other side of the equation — distribution. As an early-stage tech investor, we are all too familiar with the fallacy “build it and they will come,” so it was time to get our product in front of our target audience and, as such, I needed to think about marketing messaging, and, at the very least, a website.
This is where things got hard. ‘There’s An AI For That’ lists 55 different website builder AI tools in the market as of today, and I tried out some of the more powerful ones to see if I could get a slick design up and running, given design is a skill languishing at the bottom of my left-brained dominant self. And I just wasn’t happy with anything that was created. Enter, a human — a dear friend who can do this stuff in her sleep, and a couple of days later, we launched a website with a design I loved (including a bespoke logo) and let the AI excel in other areas — the website copy.
That brings us to today and the early days of our marketing strategy. Again, this is an area where ChatGPT has been helpful in setting the strategy and outlining the top priority areas to focus our early efforts on (namely content marketing, SEO and paid channels). In fact, this post marks the beginning of our ongoing content strategy if we choose to continue this whole experiment. ChatGPT has already provided us with titles for our next 52 blog posts to help drive traffic to our website, and it won’t take much effort to work with ChatGPT to flesh these out in detail. On the paid marketing front, in many ways like the website design tools, we appear to be in the very early stages of useable tooling and, again, I haven’t found something which takes all the pain away from someone who has limited experience in Google or LinkedIn paid ads (the two channels recommended to me by ChatGPT). I started to use ADXL.AI but wasn’t impressed with the early results and continue to look for alternatives to test out.
Reflections on AI and its Utility
It wasn’t a straightforward build from idea to production, but I got there, alone, and in a matter of weeks in my spare time. Each line of code, each integration from Stripe to Slack, was facilitated by my AI co-pilot. AI didn’t just help in building; it advised, suggesting optimal pathways, and sometimes, stopping me from going down a rabbit hole.
My advice to those still contemplating dipping their toes into the AI world, jump right in. It is much more fun rolling your sleeves up and building something, and I honestly believe the tools are now at hand to launch products and solve problems.
AI is a massive enabler for the ambitious, brave, and persistent, and I am not the only one who sees the world this way today — the team at NFX believe that $100m+ revenue businesses will be built by highly-talented AI-augmented teams, allowing for scale never seen before.
There’s an entire ocean of possibilities waiting. As we stand at the crossroads of technological evolution, my advice is simple: embrace AI. The tools are here, accessible, and affordable. It’s time to move from contemplation to action. AI has proven to be more than just a tool; it’s the co-founder you were looking for.
