Pear AI founder: We made two big mistakes
twitter.com
twitter.com
As an aside, I'm starting to get annoyed with the "we moved fast and broke things, we're smol indie hackers" defense that has been popping up more lately in various contexts. The whole incident started because they got venture capital for their fork and were in a startup accelerator for several months: that's not moving fast or being indie.
It's worse than that; this whole kerfuffle came to people's attention after the founder posted a brag tweet about leaving a $300k job at Coinbase to go YC. Only to claim ignorance about the license and everything else later.
That's in general not specific to any situation.
More specifically PearAI forked the OSS code editor Continue, which was itself funded by YC, and got YC funding for it. Also the editor they forked is itself a fork of VS Code, but is not to be confused with Void Editor, which is a third YC funded VS Code fork with AI features. It's YC funded VS Code forks with AI all the way down.
They aren't funding the viability of the product, they are funding for a take of future earnings of the morally bankrupt or the righteous will of the hopelessly naive. In either case, it's all about how effectively they can affix a leash.
Rent-seeking over value-creation.
A mission that is, "I have an AI, input is business plan, output is entire IT infrastructure" is what YC sees here. i.e. programming as a field (quite a cost center) is no longer needed, ironically. Because of that the bar is low. Honestly if anyone is reading this, if you find a very clever novel way to go after this, surely you would be funded
Are you talking about continue.dev?
That's not an editor itself, nor is if a fork of VS Code; it's an extension for VS Code (and JetBrains).
https://www.youtube.com/watch?v=X0OylwLzBQw&t=257s
Which I suppose isn't necessarily a gotcha, it's totally possible to be a user of software and figure out ways it could be improved. But the way they pitched themselves was they're building an AI code editor, whereas it seems like they're just making some changes to a free, open-source project.
If I had any guess, the reason they got accepted is from their YouTube followings, and this was essentially a marketing experiment. These guys have audiences of young aspiring developers who could be potential customers if their favorite youtuber were to start claiming they "built" an AI code editor. Not a crazy idea but certainly doesn't smell of an attempt to make an honest and good product.
I wonder if it's possible to get funding for a fork of an already funded fork that you launched recently.
Might even be an easy money glitch: every couple of years or so you fork your YC funded code editor, get YC funding for the new fork, and cruise on that money for a while.
(all of the above is said tongue in cheek)
Don't give blockchain bros any ideas...
But seriously, this has got to feel pretty shite for continue. What is their impetus to roll out new features when they know that pear can focus on only the icing on top of that milestone, inherently making continue look like a less polished version of pear, when in fact they were the innovators?
I think what we have here is an antisynergy of the incentives baked into open-source and capital speculation. It’s a recursion bug that undermines the incentive structure, and YC was foolish to be so disloyal to their partners.
I don't see many AI companies reaching billion dollar valuations and I don't see any AI company retaining billion dollar valuations in the next five to ten years.
I think the safest bet is in niches where you bring unusual or hard to obtain datasets to the table. Training data is the only moat.
people seem so caught up in founders' particular ideas at the pre-seed stage and forget that YC is investing in people - specifically: people who try. you can jokingly quip that you'd be better at forking vs code, but at the end of the day they're doing it - not "waiting for their check" - and you're making fun of them on the internet ¯\_(ツ)_/¯
Are they? Seems like based on this YC alum [0], the reality these days is otherwise.
Also, maybe consider you're spending your time baselessly defending strangers on the internet. ¯\_(ツ)_/¯
Which is of course fine but you see Dalton's Request for Startups and I've not seen only a tiny handful in the last two batches.
This makes more sense now that I think about it.
A legal wrapper will use domain specific legal expertise to structure and develop their product and functionality. It will likely have many functions that are specifically optimized for the legal use case.
If a wrapper can easily move across many domains, then the wrapper likely adds little value to the base model.
It’s like somewhere in between getting grant funding for a university project and real corporate work.
I don't get it either - why spend $2m on 4 YC companies under different group partners, creating a hunger games of sorts, instead of convincing all of them to combine their talent for $500k. Sure, egos will be bruised and they might fight over who's boss, but the chances of the final product succeeding would be much higher.
Edit: I just realized the old t-shirt slogan of GitHub was “Fork you.” Even better.
It's like saying 25 years ago that it doesn't take an expert to know that you shouldn't put your credit card on the internet.
Reading the room, people are upset about grifters, scams and fakery in general. The outrage is obviously built up over seeing our industry decline to buzzwords, bait and switch, and general enshittification. This is just one particularly clear-cut instance of opportunistic grift caught in the act. It’s like that IoT juice press – it’s not that harmful, but it’s a symbol of what people hate.
They are experienced software developers coming from $300K jobs at Coinbase. As experienced developers they know the requirements in the Apache License 2.0. As former employees at highly regulated Fintech companies like Coinbase they know better than letting ChatGPT make up new legal texts.
They are cynical manipulators hiding behind a made up fictive "careless Gen Z tech bros" image.
I mean... crypto companies are well known for being chock full of decent and honest employees who aren’t zealots or bandwagon jumpers in any way, shape or form.
Just like a lot of recently created “AI” companies.
The compensation is on the low-end when it comes to what Coinbase pays to its employees. They wronged their hires on multiple occasions with dilutions too. Many of these employees had strike prices that were above what Coinbase IPO'd at. They were relying on their options, otherwise their base pay was undermarket.
I’ve worked at a company with hundreds of devs where I caught an obvious error (a reference to a section that did not exist) in the pre-employment agreement. To imply that most devs are carefully reading documentation in general, much less legalese, is laughable.
Um.
Here are the commits from the founders:
* https://github.com/trypear/pearai-submodule/commits?author=n... (71 commits)
* https://github.com/trypear/pearai-submodule/commits?author=F... (21 commits)
This is what got funded... I leave it to folks to decide if the changes to date are meaningful.
If someone builds an electron app and gets VC funded, is that somehow unethical?
Lines of code is irrelevant. Only issue is if they are building on top of a codebase whose license they have violated.
https://github.com/trypear/pearai-submodule/commit/bf67ece35...
Just dd a promptfrom one object and p it to another.
https://github.com/trypear/pearai-submodule/commit/c59d90a4c...
change a keyboard shortcut lol
https://github.com/trypear/pearai-submodule/commit/489004ae0...
manually set font size in 3 places but remove the variable that would still be valid in 2 places???
https://github.com/trypear/pearai-submodule/commit/9bbee4c3d...
made an error message longer
Now to be fair I've never been paid 270k by coinbase so presumably these guys are a lot smarter than me, but I'm still really confused why they got funded.
You get hired and undercontribute for nine months before putting in notice after you win another interview, which is easier this time because you won one before and got the pedigree.
There is a whole generation of people who write solutions to assignments and have no idea how to create software. These commits are what happens when they don’t receive a prompt requiring a from-scratch implementation of a topological sort.
Lines of code is a proxy for "how hard would it be for someone else to do what these guys are doing but better" and the answer here is "not very".
Let's take it to the extreme - how much would you fund a "tech startup" that doesn't have any of its own code at all, but rather just has its founder manually press buttons on a third-party tool's GUI?
Note that I'm not asking whether it's ethical, or even if the service is worth it to a customer (maybe the founder really is good at pressing the right button at the right time) - but would you as a founder fund it, knowing that there's pretty much no barrier-to-entry whatsoever?
Not to mention the whole hype around AI is unbearable at this point. Seems very likely that deserving products/ideas are getting rejected because they aren't doing AI, or because their founders didn't go to Stanford. Oh well, that's just how it goes.
Unless you are from few *US* unis, you have zero to none chance of getting in, unless you have significant traction already, in which case you may as well go to any investor and they will fund you.
And if you are among those lucky few, they will fund you with the most shitty idea you have, or even without any at all.
This is the part that really bothers me unfortunately. I honestly grew up thinking that YC invested in people and ideas more than traction. If I had traction, I could raise money from anywhere. YC does invest in people without traction, but you have to be a certain type of person, from a certain background, and from certain institutions.
That's great, but just look at the profiles advertised on the front-page there: Harvard, Berkeley, MIT, etc. Nothing against them personally of course, but I really think this is the trend YC is going towards these days. It's the old school institutional prestige back, but this time it's blended together with SV startups.
A lot of YC's advice is terrible, such as their advice on finding a cofounder.
You can easily tell when they give bad advice: ask them for their dataset. When they don't offer it, consider doing the opposite of what they tell you.
Now, some of their advice is great, when it is data backed an easily testable like Jessica Livingston's "I don't know of a single case of a startup that felt they spent too much time talking to users."
https://breckyunits.com/advice.html
Now, what's a version of their "find a co-founder" advice that doesn't suck: "I don't know of a single case of a successful founder that wasn't building things all the time with many different people."
Don't find a co-founder or two, find dozens of "building buddies" who you just build shit with. Every now and then you can form a legal entity and be "co-founders", but the more important thing is you just create stuff together.
> We thought the license in the root repo wasn’t that important, so we just generated one that we thought was open.
The root repo already had an Apache license in it. If you thought it wasn't important, why replace it in the first place?
Those capital letters and correctly spelled words and full stops come off as so passive aggressive.
Otherwise it painfully looks like he tasked ChatGPT to generate his apology as well as his license.
ie, game the test, game the admission, game everything. and some folks see this as "hustle".
eventually their own process gets gamed.
gaming incentivises a culture of "pretension" - you dont know sh* but you fake it. fake it til you make it. copy code-repos, trample over licenses, whatever, who cares as long as you are getting ahead.
There are some exceptions but the typical demo day is absolutely stitched together from other peoples’ work. The difference is most hide this fact, instead of working in the clear on “open source” without understanding that it is meant to be a contribution from you to the community, because to finance bros that sounds like a joke.
The whole thing has gotten so washed out that they either thought that this is what people actually do, or they thought no one was going to look at their repo (for a coding product!).
Where it's all about the hustle, founder mode, the scene and whatever you can get away with to make money and take advantage of people.
Whatever you may think of Pear AI, or startups playing too fast and loose, IMHO this apology sounds sincere and worthwhile.
Contrast that to their comment about "100+ contributors".
It feels like typical deflection.
Also egregious is "We thought the license in the root repo wasn’t that important, so we just generated one that we thought was open." It's hard to trust someone who would think this in the first place.
Really makes you wonder what YC even looks for in a business these days. They're certainly neglecting their due diligence.
You can't "tried to be transparent about what we’d done" and "made it look like we were stealing the work" at the same time.
Either you announce yourself as "Pear, the VS Code Fork that will change the way you code", or you try to be very low key about it yet hope to retain plausible deniability when people call you on your BS.
> In summary, the apology appears mostly sincere but is tinged with some defensiveness and rationalization, which might affect how it's received, especially by those who were most upset by the initial actions.
Clearly they are using “moving fast” as an excuse because they have no moral compass
Smug self entitled YC people thing they are doing amazing things because YC says so.
I Ranked the Worst Influencer Apology Videos - https://www.youtube.com/watch?v=ZYeR7hvpTfw
> Our intention was to use the Apache 2.0 license like Continue does.
So why didn't they copy the license like they copied the rest of the code?
> We thought the license in the root repo wasn’t that important, so we just generated one that we thought was open.
Why would the license not be important? And why would they think that the one they generated was open? If they were going to blindly copy and paste from chatgpt, why wouldn't they blindly copy and paste from the original codebase instead, which is already known to be open.
Software licenses are a core part of any software company's business model, so I don't believe for a second that these YC-backed founders didn't realize what they were doing.
... With that said, Apache and all other "OSI approved" licenses are open to being exploited like this, so I don't think they're necessarily in the wrong. The apology, however, is 100% bullshit and I wouldn't trust these people with anything.
still doesnt make pear an investable idea, but the founders showed some humility/ability to read the room here.
conversely, people who dunked on -them- too much showed either their lack of knowledge on open source/startup norms, or their ability to disingenuously ignore that for internet points, and provided many opportunities to be muted.
Here is a clear as day example of a startup who were completely reckless and indifferent to (a) the principles of open source and (b) the concerns of the developer community.
And instead of criticising them you've turned the tables and now the fault is with people like me who are doing this for internet points. Instead of defending what underpins the entire software industry and has contributed so much value to the world.
Mistakes are made without intent. That's not the case here.
1) management types who have never coded love the idea that coders can be replaced or minimized by said tooling. hence hey fund any and all projects.
2) coders excited by llms reach out to the nearest projects they can relate to, namely, code assistants, test generation assistants, doc generation, and generally any automation in the coding workflow. so, you will see a lot of me-too tooling.
Venture capital is trying to fund companies that have the potential to grow their valuation many many many times over in a relatively short amount if time.
AI is obviously very powerful and has the potential to change things quickly IF it turns out it can be good at important tasks that people will pay for. If someone finds a really good use case then they can scale that with essentially no problems until the vcs are fat and happy.
Pouring money into AI garbo also feels like a hedge to investors when they see it as just funneling down to their other investments.
Together these things end up looking like an attractive vc target so we see a ton of these "companies" that are a bag of prompts on top of a different company's code/model.
Software development itself is a joy that I'm not willing to give up as a hobby.
I'm not saying "all software is bad" at all. I'm not even saying "all software companies are bad". I know that's not the case. What I am saying is that the industry overall, mostly due to the actions of SV-style companies, has become more bad than good and the general public is noticing.
And we also chose to extend the presence of gambling in society - quite literally with sports gambling - but also adjacently with crypto markets and stock market apps
Tech has always had advantages around being able to slack off if you were good since it was a white collar desk job, but you could justify that to yourself as a sort of implied retainer arrangement. So it wasn't directly grifting or stealing.
Some people get a taste of real money relatively easily (like a $200k+ salary for slacking off) and want more though, maybe because they're afraid that good situation can't last and they need maximum value extracted before they're "found out".
And we are just not seeing major downsides to grifting unless you run afoul of the law, which can be slow to prosecute if you are on a bleeding edge like AI (like it was for crypto).
Have you worked at a big tech company? Because no you don't.
Similarly, with their extension repo which they forked from Continue, they now have two LICENSE files, one from Continue, one from PearAI, both Apache. This could be argued to be almost correct, but it would be clearer if they did it correctly and had one LICENSE file there with both copyright clauses from both PearAI and Continue.
Additionally they don't seem to use any contributor agreements, so people who contribute to their repos will maintain their own personal copyrights. That means that the copyright clauses on the licenses are technically invalid as the repos contain works from various individuals who have not given their permission to license their works in any way.
Disregarding the legal aspects, what they are doing is morally clearly wrong as they are crudely exploiting the works of others without contributing back to them in any way.
I instantly disliked them from their announcement. Huge crypto bro scam energy. Even that aside I decided to give them a chance and look at their product. Oh its just a combined fork of two other products. Well even so I gave it a chance and installed it. Maybe they are just marketing? The product is incredibility underwhelming, just another bland AI "Me too" clone product.
Then I see that they actually are crypto bros from coinbase....They have been drinking from the kool-aid mines of crypto too long their compass is gone. This is just them trying to shamelessly "value extract" one of the last bastions of real production. What creeps.
Plus all the other drama they have been involved in ... They are the exact antithesis of the startup world. Poison in the well. I hope they fail because they deserve it and this community deserves better than being exploited.
His original tweet was using the exact same tactics as his video titling with bold words shouting "I QUIT MY $270,000/YEAR JOB AT TRENDY TECH COMPANY" that I think a lot of people outside of aspiring first-year CS students aren't falling for after we've been exposed to so much BS grifting out of silicon valley the past decade+.
I do. I don't believe anyone working in the crypto industry is anything other than a sociopath or incredibly naive. They clearly are not naive since they are pumping their own product with deliberate and IMO false intent. How many times do we have to watch bad actors with their hands in the cookie jar before its a pattern?
>so much BS grifting out of silicon valley the past decade
Yet here they are still getting money. The real downer is that every one of these people are draining limited money that could have gone to someone genuinely trying to do something.
I also believe this could be an inciting incident that leads them to think differently and approach things differently. They could also go right back to their old ways. But I think this is a pretty good day-one response.
_Generated_? Oh, yeah, that’ll go well.
Of all the comically inappropriate uses of LLMs I’ve heard of, _writing legal documents_ has to almost number one, behind perhaps only medical use.
This apology is even misleading of what they actually did. They didn't fork they copied and they changed the license. Fork properly keep attribution to all the developers who have worked on continue; and keep the license..
Maybe there's a billion dollar company in that or a sustainable business model because consumer hardwar may never be able to run LLMs.
But I really don't know because otherwise I'm sure pretty stable, serious ideas with proven demand would be rejected outright.
Rather, you now have company in your portfolio with paying customers right off the bat that it would not have otherwise or would have taken longer than usual.
I suppose they figure they can just buy it out in court appeals or something if someone were to take a grievance up with that.
How does Graham, et al justify this lackadaisical attitude... "You win some you lose some"? "It all comes out in the wash"? I guess if you got masses of money to throw away you don't particular care about the legal ramifications? It's just an entry in the books.
How did YC not smell that as a flag, and why should anyone TRUST YC after such a misstep? Or other VCs, not to dish just on YC... How many other situations like this are out there?
Are they going to work to verify licenses with all their projects? Or just wait til disaster erupts and someone finds out they done messed up (oh I know the answer based on the SV motto "ask forgiveness instead of permission").
I am slightly snarky with my attitude, but this is serious shit, and it matters, because the social contract matters, SV Homo Superior Morality be damned. This is the kind of attitude that leads to bank collapses, because they're so busy huffing greed and damn the consequences. In this case, it's obviously not nearly as catastrophic, but it gets tiring to see this attitude that they get to reap benefits while externalities? Those are for the little people to pay.
I am not a lawyer, but I guarantee you running this strategy past any lawyer will at best get you laughed out of the room.
YC has a long history of funding companies that blatantly break laws such as AirBnB and Uber, so I wouldn't expect they'd be bothered by this.
>"Only A players make it in this game"
A license like MIT explicitly allows commercial use.
If you don’t want other people to make money, don’t use an open license.
If you want to make money, then charge people for it.
Choose a license that aligns with your needs. Be prepared to sue if someone violates the license.
Why would you believe anything this person says after that? Default assumption #1 is any writing they output is an LLM product and insincere. Assumption #2 their actions are taken with little thought or intentionality.
He deleted the tweet, so link with pic (I don’t endorse the generational dig in this tweet): https://x.com/anothercohen/status/1840515897804623882
This one line tells me that's an outfit that should be avoided entirely. It's either unfathomable incompetence, or a strong aversion to doing things properly. Either way, it says nothing good.
Airbnb and Uber succeeded by simply pouring cash into funding legal challenges, and waiting for cities to give up. OpenAI can't chase after Middle Eastern cash fast enough. Musk and Thiel are openly backing the most brazenly corrupt presidential candidate in US history.
The move that dials the dumbassery to eleven is using it as a defence. On Twitter. Like, Exhibit A for any lawsuit that company is ever in will be this tweet: it demonstrates a proud disrespect for law and contracts. That’s high-proof mens rea if I’ve ever seen any.
I was given three pieces of advice on starting up my own business and they were good:
1) Get a lawyer
2) Get an accountant
3) Listen to every word they have to say
Because legal naïvete is common and isn't a good predictor of founder ineptitude. Is it better to be legally savvy? Of course. But thinking you can wing it with a license agreement because it's boring and unfamiliar and you're rather focus on building your product is understandable. (In some cases, it might even be the right call.)
I'll stand by it. Most Americans are legally naïve. Most founders are, too. A start-up has to make trade-offs--it's far from clear such licensing agreements are make-or-break at the angel or seed stage.
Once that's pointed out to you, you hopefully either hire a lawyer or mitigate your legal cross section until you can afford one, e.g. by using template license agreements instead of rolling your own. But until you've been given that feedback, it's not some bizarre conclusion someone can come to. Most small businesses, for example, are formed perfectly well without much legal sophistication.
However, using templates is a far cry from asking an LLM to write your agreement and assuming it won't hallucinate something that's going to put you in legal jeopardy.
Sure. Not disputing it isn’t incredibly stupid. But if, on having that pointed out, the founders had admitted it was a dumb thing to do—they had acted thoughtlessly and recognised as much—I’d be willing to forgive them the error. Hence, forgivable.
ChatGPT'ed license signals to me that whoever does that thinks law is not a serious matter rather BS that can be generated via BS generator.
Reasonable people can both respect the law and think license agreements are meaningless gibberish.
They're wrong. But not uncommonly--and I'd argue, not unreasonably--so.
Not realising you're legally naive is beyond stupid. Extraordinarily stupid.
I would assume it would be just an email away for a YC statup.
The "we busy building rn can't be bothered with legal" is especially hilarious given the paltry amount of coding they have actually done so far.
Attribution is one of the most basic precepts of decency. Not even "open source" or "free software", just basic decency. Mistakes happen and that's okay, but being all derisive about it initially, and then trying to spin that as "we learned about licensing" after people call you out on it is hard to take as genuine.
A genuine good faith response would have been "oops, what a silly embarrassing mistake" and then spending all of 30 seconds fixing it.
These people "like" open source only as means to extract value. They are only "part of the community" when it suits them. Nihilistic cryptobros considering everything that's not nailed down as a wankdoll to be abused and extract value from – who would have expected?
The detrimental English usage however is.
Is it supposed to hurt using those words?
1. Maximize attention on social media by being super obnoxious and arrogant ("dawg I ChatGPT’d the license")
2. 1 day later while the chatter is still going, write a mea culpa and take on the poor victim role ("grew up in a single mother household on government subsidies")
3. --> Repair most of the reputational damage but keep all the attention.
None of this is illegal, but it's exploiting a system of mutual trust and I wouldn't want to live in a world where everybody acted like that.
Can I get more YC funding by forking Redis, adding OpenAI, and calling it an AI In Memory Cache?
No. But, maybe if you call it RediAI
Also OpenAI was founded before the transformer paper and spent most of their early years focused on RL style research and products - just look at what they published pre 2019.
99% of the people here would make this mistake.
and to be accused of taking something if you are giving attribution would just be absurd
OSS community has some lame and immature aspects
These guys might have gotten defensive but its obvious how to communicate, to me
Doubtful. That said, some may.
> guys might have gotten defensive
Now you've lost me. It's fine to be legally naïve. But you should have the self awareness to know you're winging it. When someone gives you feedback, especially for free, it's a damning personality trait for the first reaction to be petulance.
You know what's even easier than having an LLM generate your license? Copying one that already exists. I know a cool one, it's called Apache 2.0
This person took on (a little) additional work to do this. Since they used Apache 2.0 in one of their repos, they seem to grasp the concept of public licenses. If this truly was pure incompetence, it would have really been major.
> We thought the license in the root repo wasn’t that important, so we just generated one that we thought was open.
No part of this sentence makes any sense to anyone remotely informed about licensing.
the reaction was attributing it to malice, when it obviously was incompetence
that should have been 1 tweet or github issue opened asking them to change it, and it would have been corrected or not. firestorm if they didnt take it seriously
instead of whatever all this drama was pre-emptively
Unless the quality of HN has dropped to such catastrophically low levels that TikTok comments section infested with bots is higher, no.
Seriously, this is ignorance at its finest.
99% of people here would… have spicy autocorrect generate some legalistic-looking nonsense, for, based on the above, _absolutely no reason at all_ (given that they’re claiming they believed that it was still under Apache 2.0)?
I mean, this place is bad, but not that bad.
Cmon guy