RIP.
If you’re not leveraging AI internally in your startup, you’ll be left behind by startups that are. We all know how powerful an LLM like GPT-4 can be, especially when pointed at something unstructured or nuanced that would typically take a human’s touch. Just look at GitHub copilot. Once an engineer starts using that, there’s no going back. It’s akin to not having a microwave in your house. You probably won’t cook all your meals or even the best meals with it, but damn, is it convenient and useful for those little things. Almost everyone can get one due to how affordable and accessible they are too.
The value your startup unlocks by adopting AI tools is too much to ignore, especially because your startup could die without them today.
That same convenience that allows you to AI-ify
your startup is why AI alone is not an advantage. Whether you’re using AI internally or as some feature/experience in the app for your customers doesn't matter.
Everyone’s doing it.
Does your choice of front-end framework or database give you a powerful enough advantage to move faster than the rest of the market? Probably not at this point. It’s because anyone can use those as well. They are a commodity, table stakes. Imagine building something today of the scale of Netflix from scratch, with no frameworks, no tools. That would be a horrible time for anyone working on the product. It’s almost unseen for that to happen because of how many tools and frameworks exist to help you build your product. Using any of those tools is not an advantage, though. React, Django, or AWS will not give you a leg up to win in this market. Already this is ringing true for AI tools as well.
AI is an accelerant, not a competitive advantage.
The answer is yes if you’re a founder wondering if you should incorporate AI into your processes. For a while, privacy was a big reason why a team wouldn’t use AI internally, but that is also changing with new open-source models like Llama 2 that are on par with what OpenAI is offering. There is no reason not to augment your team with AI to move with less friction.
Fomo features.
It's not so straightforward when thinking about adding new features to your app that expose AI to your customers. I believe the best interfaces for customers interacting with LLMs are hidden and not like a chat. You can drop a GPT-powered chatbot on your site or app to allow users to “chat” to their data. Is that adding value, though? Were your customers going out the door because you lacked such a thing? For most, I’d say no.
So please don’t just grab the GPT-4 API and start introducing AI in your app in places where it’s not valuable. Stay disciplined and true to your product process.
Sign out.
So if you’re a tech startup, explore ways to use AI internally. To help engineers write RegExs, they’ll f*ck up anyway, assist your marketing team with some first draft concepts, and step in to help designers with art as placeholders in their designs. Unless your product is AI itself or you are working on infra or a foundational model, don’t be so quick to create some AI add-ons in your product for the clout.