WHY THIS MATTERS IN BRIEF
If you build your company on top of another company’s products minor changes to those products can destroy your business overnight.
Love the Exponential Future? Join our XPotential Community, future proof yourself with courses from XPotential University, read about exponential tech and trends, connect, watch a keynote, or browse my blog.
I’ve seen this many times before – a company builds a product, start ups see an opportunity to add extra value and build something on top of it then BAM! an update wipes out their value proposition and death and de-valuation follow swiftly afterwards.
Start ups that rely on another company’s product to do what they do though don’t just die because of a product update. An example of this was when Twitter, now X, for example, suddenly changed the cost of accessing it’s API from a few tens of thousands of dollars a year to over $250,000 – a move that wiped out almost every company that relied on accessing Twitter to analyse or publish data.
The message? Beware of who rely on … Now OpenAI, a company which lots of so called wrapper startups rely on, might have also just killed a bunch of them with a new minor update as well after Sam Altman’s company introduced a new beta feature for ChatGPT subscribers on Sunday which allows them to upload PDFs. The chatbot analyzes the files and allows users to start asking questions about them.
Anyone who has ever tried scrolling through hundreds of pages of PDFs to pull out useful data and synthesise information will know just how tedious it can be. That makes the update a pretty nifty one for ChatGPT Plus users.
The update also creates a wider problem.
Some startups have built their businesses around ChatGPT’s lack of interactivity with PDFs. Now that ChatGPT can interact with them, what can these other startups offer that ChatGPT can’t?
The simple answer may be a whole lot of nothing.
Sahar Mor, product lead at payments giant Stripe, wrote on LinkedIn that “OpenAI just executed a move that will wipe out dozens of AI companies.” Specifically, he referred to “Wrapper Startups.”
These are essentially start ups that “wrap” themselves around an API like ChatGPT, using the core technology underneath the chatbot to offer some sort of service that isn’t available directly from that API.
It isn’t necessarily the case that founders who build AI wrapper businesses are doing this to exploit a weakness in ChatGPT. In March, OpenAI flung the doors open for developers to integrate ChatGPT into their apps and products.
That means startups offering PDF functions aren’t the only wrappers in town. There’s a whole bunch of them out there offering all kinds of uses.
Perhaps the most prominent wrapper to date has been Jasper AI, which went into 2023 with a $1.5 billion valuation, with backing from big VC players such as Coatue and Bessemer Venture Partners.
What did it do to get that lofty valuation? It created an “AI Co-Pilot” specific to the needs of enterprise marketing teams by wrapping around OpenAI’s GPT-4 model.
It’s an idea that seems to be struggling, with the company cutting its internal valuation, The Information reported. It also announced layoffs in July.
Startups that offered user interactions with PDFs by wrapping around tools like ChatGPT may wonder if they face a similar fate.
Here’s another example. In May, data scientist Alex Reibman announced ChatOCR, a ChatGPT plugin that “reads text from PDFs, including scans and handwriting.”
Following the ChatGPT update at the weekend, he ran a poll on X that asked users what would happen to his plugin “now that ChatGPT has built in PDF processing.” Out of 210 respondents, 72.4% said they expected the plugin to “see less usage.”
The key takeaway is that without a moat to make them significantly different from competitors, startups risk being plundered so customers and investors alike need to be much more careful about the people they partner up with and invest in.
