A recent controversy emerged when Y Combinator startup Pear AI appeared to duplicate Continue's AI-powered IDE concept.
Same idea, different name. This wasn't a scandal. It was a spotlight on a fascinating question:
How often does this happen? And more importantly, how can we tell? Let's find out.
As visually fluid as it is, it's also worth taking a look at disclaimer about the limitations of our research
Y Combinator's portfolio of 4,936 companies provides the perfect testing ground for AlphaLens, our newly developed AI search technology. With data spanning 20 years and 39 batches, this rich sample set lets us demonstrate how AlphaLens can identify and surface similar companies through sophisticated product data analysis..
Take's company description: Economic infrastructure of the internet
is sufficiently vague, that, when you run it through a sentence embedding model along with all the other YC company descriptions...
O.S Infrastructure for the Financial Internet.
The shipping infrastructure of the internet..
Cross border Payments infrastructure..
Takeaway #1: The true significance of a company's offerings, often becomes obscured by the generic nature of a standard company description.
A clearer picture emerges, showing how companies actually compete through their respective products' USP, features, pricing and more. While this detailed information is incredibly hard to gather at scale, that's exactly what AlphaLens has done. By organizing millions of products into high-refresh structured data, we can show you precisely how Stripe's offerings stack up against competitors.
Hover on the any of Stripe's products to see what we are talking about
"A subscription and recurring billing platform designed to help businesses manage recurring payments, automate invoices, and handle tax calculations." By analyzing at the product level rather than company descriptions, AlphaLens reveals remarkably relevant competitors. Hover below to explore the matches:
The results we are getting at the product level are far more relevant. and if we zoom back out to the company level...
It is far more detailed and intricate than what we started with at the company level.
So if it works well for one company, why not...
Stochastic (t-SNE) dimensionality reduction allows us to plot these complex product descriptions a 2d plain, which whilst not perfect, is helpful for showing clusters.
The darker orange areas are where the density of products is higher - hover and click to explore.
The dimensionality reduction reveals distinct semantic clustering of product domains, with clear evidence of thematic proximity between technologically adjacent sectors.
While we've annotated only a few prominent clusters (including Biotech, AI Code Editor, and Crypto), the underlying data reveals many more subtle groupings across the startup landscape.
To cut through the noise, we've highlighted a few more clusters that are noteworthy in their density. You can leverage the dropdown and/or the legend to change the view.
The two companies that kicked this all off, Continue and Pear AI are not the only ones trying to build an AI powered IDE.
With some of YCs largest winners: Amplitude Mixpanel Posthog and June all vying for the same customers.
If we gather the clusters and plot them against Batch year, we start to see where trends emerge and fall.
Fundamental business tools like e-commerce and analytics saw a rise in bets in the early years.
The middle period (2012-2016) saw a shift toward specialized B2B point solutions
2020 onwards marks a dramatic turn toward AI, with dense clustering across workflow automation, code editing, legal assistance, and CRM.
Quite a few it seems
Of particular note, the target audience in most of these are restaurants / cafes / liquor stores. Digging deeper still we can start to see which group partners backed them (color dots) and which intake.
Strikingly, Craft OS and Santé both:
are POS systems
serve similar customers
are part of the Summer 23 Batch
share same group partner
Here's Expense Management Software
Strikingly, Alaan and Bujeti also both:
are expense management platforms
serve similar customers, albeit UAE and Africa
are part of the Winter 23 Batch
share same group partner
received huge attention in W23 with Blyss, CodeComplete, Cosine getting backing in the Winter batch, and Continue, and CodeStory both enrolled in the summer batch.
Among those, 3 of them share the same group partner, and 2 of them were part of the same batch... awkward.
As you can map it against batches (colored dots) and see what bets are unique and which ones are hedged
Interestingly the large yellow cluster contains both Abel Police (S24) and Flock Safety (S17) both work with Police bodycams, to render the evidence capturing and paperwork more efficient for law enforcement. Albeit through different means.
Feel free to explore other group partners' tracks with the dropdown.
But if you came this far, you probably want the data, right?
Just ping us on contact (at) deckmatch (dot) com and we'll share it with you.
It's worth noting that this is just a tiny fraction (0.2%) of the 6 million products available on Alphalens but we hoped it was insightful and fun to read.
But, if you want access to the full platform...
You just need to request early access to our search platform
Or if you are an operator and just want to map out your own space, as in the video -> get started for free today on app.alphalens.com
Thanks for sticking through to the end, hope it was worth your time!