Did OpenAI just kill a few AI startups?
A few, and what does it say about the future?
Yesterday, OpenAI released PDF reading features as a part of ChatGPT.
While AI is doing fascinating things that were not considered possible some time ago, reading a PDF and analysing its contents wasn’t the first thing I thought of when OpenAI released GPT 3.5 Turbo.
It is like meeting the Wright Brothers, discovering the possibility of air travel and coming out excited about the future of air-conditioners. Certainly, air-conditioners have been helped along by the airline industry.. but.
Photo by Geometric Photography on Unsplash
But many companies did and rightly so, and created a super useful PDF reading tool. It is especially beloved among lawyers who trade in PDFs as often as they do Docx files.
But the startups that lead the space now find themselves jolted. Which brings us to the question:
What kind of startups are difficult for OpenAI to kill?
Here are some ideas:
AI + specialised expertise in an unrelated area makes its more challenging for another set of AI/ML engineers to just come in and copy what you did. Or for that matter - a company that has raised more money or OpenAI/Google using their own teams. Software needs to meet another kind of expertise, such as deep integration into a business vertical, or complexity that you are better positioned to handle than any other tech team. That’s a moat.
Localisation and being closer to the problem. Use cases that are specific enough not to be global but complex enough to demand deep focus and expertise. This means knowing the Indian and Chinese customer better, or knowing architects better. Your business is not a feature game, you are close to what works and know how to build better.
I believe many of the big leaps in productivity resulting from the use of AI will occur in industries and startups that: (a) are cross-disciplinary; (b) have had a crack at the problems attempted at being solved, without or pre-AI; and (c) are able to quickly build or expand into the best that AI has to offer.
What do you think?
Until next time.