Hacker News new | ask | show | jobs
by AndrewKemendo 2395 days ago
I empathize completely with this team because my company was in a very similar situation in 2018. We had two very technical computer vision products, had found traction and a growing enterprise user base but revenues didn't grow fast enough and all of the major companies were entering the market.

We were lucky enough to find an acquisition off ramp last year but all of the feels are the same.

The big takeaway I learned is, if your differentiating product/service could be classified as a feature (which most ML or CV products are) inside a platform or application, you'll be run over by the major platforms who rebuild your product/services inside their platform.

Your only hope is that your team/data/IP is so far ahead and the acquiring company can't build what you're doing in-house more cheaply than what you're willing to sell for. Unfortunately it seems like there are fewer and fewer cases where major players can't rebuild your work more cheaply.

Second, it's excruciatingly difficult to prove the value of your product/service to a potential acquirer because you don't know their metrics, and if you do, you don't know their acquisition strategy. We did an intensive integration of our product with a Fortune 50 retailer, and based on their own numbers showed (using their own A/B tests) that our service provided a statistically significant lift in a core metric that they cared about, in this case paid conversions. Their CEO even talked about it at a public summit. However their acquisitions strategy didn't include small companies that aren't major strategic partners (Only >$200M+ acquisitions).

The worst part here is that, the founders (like I was) are absolutely in love with the technology and how amazing it is. The problem is, from a business perspective, that basically doesn't matter. You could be doing the most amazing work in NLP token inference, but if the product doesn't fit perfectly as an acquisition and it's not so compelling as to build a huge platform around, it's probably going to fail.

I wish it weren't the case, but it leaves me questioning what the value of doing really hard technology is as a startup. It seems clear that the most financially successful startups aren't solving fundamentally hard technology problems until they get to scaling something with broad product market fit.

7 comments

> ... it leaves me questioning what the value of doing really hard technology is as a startup

Totally agree with this sentiment. I think this is why most "hard technology" problems are left to huge R&D departments or the government. Both of which aren't particularly nimble or profitable. There are a few notable exceptions (Oculus comes to mind), but most unicorns don't generally deal with solving tough problems. It's mostly about product-market fit and the balance sheets.

That tech isn't left to them it's just Survivor bias. Those teams fail at hard tech all the time. They're just still here
I completely agree with all of your points, having built a Face Recognition API in the past at Lambda. If it’s a feature, high willingness to pay customers will just build in house. Low willingness to pay customers will want to use it for free or not enough money to keep the lights on. It’s difficult to build an API business with pure CV/AI/ML for that reason.

I know this is unrelated, but I want to thank you for your work on Kessel Run. It’s inspiring to meet folks in the DOD working hard on making software work for our national defense. I ran into some of your colleagues and was very impressed and excited about that program.

I appreciate you saying that!

We're really working hard at changing many different aspects of the DoD all at once, from culture to acquisitions and are starting to see some great fruits but there's still a lot of work to be done.

Agreed but be cautious in the application and don't get jaded.

Microsoft DOS started as a feature of IBM PCs, and Google started as a feature of Yahoo.

Like "tech in search of a problem" IMHO "feature vs product vs company" needs to be applied case by case.

> and Google started as a feature of Yahoo.

The DOS point is correct in spirit. Google however did not in any regard start as a feature of Yahoo, it was an entirely stand-alone search engine for nearly two years prior to the Yahoo deal. Just one year after launching publicly it was handling several million searches per day directly on their own site. Google had already become very famous, with people going directly to Google's site to use their search engine before Yahoo signed a deal to use them to power their portal search. Just prior to the Yahoo deal, Google was routinely handling 15+ million seaches per day - a large figure at the time.

Here is a Salon article from December 1998 noting Google's superiority and promoting google.com specifically:

https://www.salon.com/1998/12/21/straight_44/

Let me share another aspect, the Chicisimo and their company should go down. The whole idea is terrible, instead diversifying between people, they are streamlining wear and if they succeed everyone will look the same. As happened with pop music after ai generators were involved... some things shouldn't be touched by machine.
I have a hard time understanding the value that Chicisimo provides to anyone. The people who might actually use the app are people for whom fashion is not a chore, but a hobby.
Would you say that building technology with the hope of being acquired is probably a bad strategy? It seems to me it should be safer to try to build some kind of product/service using that technology, that way you have at least your product as a revenue channel, and your product can make a business case for the technology you're developing as well (as well as server to increase awareness).
I hope so. Building for acquisition strikes me as studying for the exam.

Resources may be better spent checking the boxes of a potential acquirer instead of working on real problems except if and when they intersect. Like valuing people with acquisition experience from the other side more than the producers of tech or solving a real problem.

"Serial Founders" with multiple exits seem to do exactly that, but it looks like alchemy to me, and takes a certain type of person to do it.

In the ~13 years I've been doing startups I've only run into a few of these people who can pretty consistently build companies for acquisition, and I'm not really sure if it's just survivor bias in a growth market or they really have some secret formula.

Hey AndrewKemendo, such a great comment, so many points to think about. Thanks for sharing!
A takeaway seems to be: if you're going to use AI in product, let the AI power a whole new type of product, not use it as a feature in an existing one.