Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Api.ai is joining Google (api.ai)
286 points by agrothberg on Sept 19, 2016 | hide | past | favorite | 153 comments


I'm assuming Google doesn't need any of the company's business, and that it is being acquired for its talent.

So, my real question: how much, on average, does a founder get as part of one of these acquihires? And does s/he actually, y'know, have to do any work after getting paid?


> And does s/he actually, y'know, have to do any work after getting paid?

Yes, they're called "golden handcuffs". You basically get a compensation package so good that it'd be stupid to leave. Usually they try to get you to stay 3-5 but YMMV.


And what of the ones that don't get brought on? Do they get anything for their equity typically? For key employees i can see them sticking around, but early hire support staff that may not be needed could still have considerable equity.


I was part of a fairly large acquisition by a big player. Support types got handcuffs of about $20k over 4 years. Engineers got anywhere from $150k/4 years to $600k/4 years. Some got even more.

This was all independent of equity. Any vested equity was cashed out. Unvested equity converted to equivalent shares in the new company, with the same vesting schedule as before.


Most startups use double trigger acceleration for vesting. The first trigger was the acquisition and the second was the termination (not offered a position). So they would vest their entire equity grant.

Of course that implies that the equity is worth anything. Most acquihires are at low multiples so the VCs tend to take the entire price, leaving founders and employees with nothing.


Even in low multiples typically the VCs creates a decent incentive for the founders to stay at the acquired company (otherwise the acquihire will fail), but they aren't getting filthy rich.


I would be curious about "most". My n=1 experience didn't have acceleration terms. In fact, all of my remaining unvested ISO's vanished at acquisition ...


Interesting. And I assume that there's a lot of fudge room on the valuation as well with more of it being shifted towards hiring bonuses for those kept on?


I like to get a right of first refusal for cofounder share sales so I can block low valuations in an acquihire where you're not getting hired.


If the founders have done their job correctly, it sounds like shotgun agreements will cause the unacquired employees' equity to vest immediately.


Aren't Google acquihire offers conditional on passing the usual Google employee screening process?

I ask this not to be negative, but out of serious curiosity.

Say your employer is bought by Google - are you axed because you slept poorly and couldn't prove Fermat's Last Theorem on the whiteboard in 15 minutes?


I don't know about Google, but I've heard of this happening during other acquisitions. A friend of mine with a Ph.D. in CS worked at a startup that was just acquired. He was responsible for the a lot of the tech in the platform they were acquiring, yet since he didn't pass their algorithms heavy interview they didn't want to hire him. He ended up working for one of the other companies interested in buying the startup.


Lol wut? This can't be serious. Can you please name this company so everyone can avoid them?


Snapchat does this when they try to acquire companies.


Hilarious. We've reached a place where a doctorate in CS is an insufficient qualification to delete a fucking file.


I have heard that Facebook also has a rigorous screening process for acquihires too, and people get sent away althought with a generous severance package.


IMO the pendulum swings both ways. I'd bet money many a PHD could not delete a file properly.


This is such a toxic attitude.

The industry complains about being unable to find enough programming talent, while dismissing entire swathes of highly trained, highly educated people for investing in themselves.


From what I've seen, people with PhDs in CS are less competent as devs than devs without PhDs.


What a PhD in CS teaches you is how to do research in CS. This skill may or may not be valuable for a specific company and may or may not intersect with raw dev chops. Production Dev is a distinct skill from CS research.


As a dev who has both interviewed and worked with several dozen PhDs and ABDs, the intersection of skills required to both be a great developer and complete (or nearly complete) a PhD are nearly zero.


That doesn't make sense at all. I do not see why a PhD would disproportionally filter out people with talent for coding, so that the graduates even after years of industry experience afterwards would still be unable to be great at coding.


No, but not having a PhD filters out people without a talent for coding. Same with not having a college degree at all - it's not that going to college makes you worse at coding, but that getting a coding job without a college degree requires you to be better at coding.


I'm speaking anecdotally, and there are many ways my sample may be biased. (Also, maybe bad PhD devs make sure everyone knows they have PhDs, while competent devs never mention it?).

OTOH, there could be something in culture and personality that pushes lesser devs into getting PhDs.


the assumption here is that people can only learn at adequately fast rates before earning their phd? aka, isnt the expectation that if you can earn a phd in at least a highly related industry you should be able to pick up just about anything. Are they going to know how to wrangle a huge dev ops stack into submission, or pump data through some realtime frame work to maximize throughput or setup paxos properly from day 0, fine. but neither does any other entry level engineer and these guys can either learn it or be placed on a different team within the company that does deeper research that would make use of their phd...


Horses for courses. If you to do R&D, go with PhDs, if you want D&D, go with devs.


> if you want D&D, go with devs.

I'll go with a mage, thief, and warrior instead ;)


I'll just say it starts with a "P" and ends in "interest"


Google does this too. There is exceptions but they are rare.


One of my former colleagues failed to pass his interview after his company was acquihired by Google. He had designed and contributed a decent portion of the core of the product. One of his projects for the company had about 50 stars on github.

n=1


I am terribly sorry. While looking up information about the company to be able to answer the reply, I couldn't find any news about its acquisition by Google. It was shuttered by its parent company and I presume a portion of the company must have been sold to Google if my colleague was being accurate. I can't edit or delete the comment. If I'm unvoted/downvoted then I certainly understand and agree!

More info: he is a director for a well-used standard in the IETF, so he's no slouch.


> One of my former colleagues failed to pass his interview after his company was acquihired by Google

amazing: they spend n million dollars just to get that talented individual, and then just throw it all away. My question is how is this expense reported/justified and who gets the difference? What about shareholder value?


They have many talented individuals who could probably figure out what does what with the code they just bought


not to troll or anything but I find it utterly stupid that this is done. Also is it a big deal to have 1000 stars on a github project you run? I am curious because I do not know what the general perception is of github projects, I mean how much do they matter? I do not have a formal training on algorithms and I am studying machine learning now. Also I have a project having 200 something stars and one having 1000+ stars


1000+ stars is a pretty big deal! I imagine you could leverage this to gain software development job interviews if you were looking.

Do they matter? It is just one of many metrics of how useful a repository is. I think it's one of the most public metrics that can be distilled to a single factoid on HN. There are absolutely better metrics to judge the quality of software.


Well, I am not sure I can use it, since it is a book :-)

https://github.com/thewhitetulip/web-dev-golang-anti-textboo...

Also I am not in US.


teaching people to build web apps, especially without a framework in a language like Go, should buy you tons of leverage.


I wasn't aware of it, thank you :-)


from what I've seen at Google specifically, when hiring for team talent, acquihires are typically performed in a binary fashion -- they take everybody who wants to go, without going through the regular screening process.

n = 1


Speaking from personal experience, that's not always true.

When the company I worked for was acquired for team talent, they interviewed all but a handful of the engineers. Around half were given full time offers. Another group were given fixed-term offers and the remaining engineers didn't make it in.

I was given a fixed-term offer. Several months and many interviews later, I made it in as a full time employee. I was never outright rejected, but due to circumstances involving my contract and project work, I probably took around 9-12 Google interviews over that time.

They aren't fun. But in the end, I'd say it was worth it.


I doubt this is always true. I've been involved in acqui{hire,sition} due diligence for a number of startup at a couple of different (large) internet companies. My experience has been that the terms can vary quite a bit, both in the set of folks given an offer (all, all or none, some) and in valuation relative to average developer comp at the acquirer.


After reading your comment multiple times, I have no idea what you said.


I believe anaskar is saying that, for the one instance they've seen, Google either takes on the entire team or no one, and they don't do the normal Google interview.


And he may be right for n=1, but I know of n>1 instances where he's incorrect and whole teams were not taken. I even know of one instance where the CEO/founder was non-technical and they refused to allow him to remain in charge of the product post-acquisition. The direct of product, who was a Google SWE before the acquisition, took over.


I read it as they take the employees, if they want to join, without any additional screenings/interviews. I assume n=1 means that's OP's sample size.


His answer is 'no they do not have to pass the normal Google screening'


> I ask this not to be negative, but out of serious curiosity.

And yet you referred to it as "BS" and provided a sarcastic example in the same post. I'd say your attempt not to be negative and to sound serious failed.


The term "BS" should appear far more often in "serious" discourse. Far too much of what is currently being passed around as "serious" is, in fact, BS.

And that wasn't sarcasm!


Yes, you are correct (generally speaking).


Google's "Bachelor of Science" interview screening process? I'd assume that they would have an interview that would be appropriate to the degree they obtained. I'd also be surprised if Fermat's last theorem would be involved in any way - that's more of a math problem, and isn't relevant to the work of most software engineers.


BS = bullshit in this case


I'm gonna assume that poster knew that and was being a smartass.


Actually, I'm not so sure its a acquihire. Google has very recently announced a bunch of cloud machine learning products:

https://cloud.google.com/products/machine-learning/

* Cloud Vision API - Upload an image, and it will tell you what is in it

* Cloud Speech API - Upload a wav and it will transcribe it

* Natural Language API - Parse sentences, extracts entities, sentiment and syntax

* Cloud Translate API - Translate from one language to another.

Api.ai would make a great bullet point to add to the end of that list!


From what I know, Google's NLP service does not include intent extraction. Intent extraction is essential for voice command interfaces, such as those in Siri and Google Now. But I don't see why they would need to buy a startup for that, when they can do it themselves even better.


What about the possibility of Google being scared of being outrun or innovated out of its position in the AI race? Google does not have a monopoly on smart people.


"Remind me ______ in two hours"

One day Siri kept hearing that as "into hours" and I almost lost it.


When this happens to product X I feel an instinct of commiseration for the users of product X.


Users honestly should know better by now.


...should know better than to use products and services online?

Your comment is as toxic as the acquihire culture itself, and only makes users feel less "part of the internet" when companies treat them as disposable.


It's a cynical comment, but one that (sadly) may serve users well.

It's like the "assume your equity is worth zero" heuristic that many of us (including myself) employ when advising friends. It's not strictly accurate, and if taken truly literally it may poison the well for the startup ecosystem, but following this heuristic would prevent painful events for many employees.


Not toxic at all. It's a very accurate comment.

Having been in numerous situations like this, I actually now "know better" than to rely on some 3rd party API that's not easily portable.

That's probably why most startups that provide infrastructure as a service are built on open source (using open standards) from day one. Otherwise good luck getting any traction


Should know better than to rely on and pay for a service without some kind of assurance that it won't simply be switched off with little warning. Maybe expect a bond big enough to keep the service running in its current form for N years?

It's better for potential users to be aware of how often this happens, than go on in ignorance of it.


Should know better than to be locked into products and services online. Buying a hairdryer from Amazon is one thing, as the hairdryer is 'immutable' but relying on a product for a workflow at work or your day to day life and having it pulled is a danger. That's why I'd use Jira or MS Word over something offered by shiny new startup. Even better use markdown files on your local drive. Back it up at Dropbox sure but make sure it is backed up somewhere else too(!).


> over something offered by shiny new startup

It doesn't have to be a startup. Google kills projects too.


The buyer typically pays $1M per engineer they keep. What the founders get depends on how much they were diluted by funding rounds.


What do the founders typically do within the acquiring corp after the acquisition? Do they have any agency? Or are they just another set of middle managers for the meat grinder?


From my admittedly limited experience, they try to fight the good fight for a little while before being defeated by the corporate bureaucracy and then sit around waiting for their vesting to end.


It's unusual for an acquihire to have real agency in a big company. It's usually the middle management route - or worse - the limbo of having nothing real to do like the parody scenario in the Silicon Valley tv show.


I was at a startup 15 years ago where the "nothing real to do" scenario played out to the letter.

They were bought by Gemstar-TV Guide, who moved the team to their office and then apparently forgot they'd bought the company. Our team pretty much had no duties other than showing up to "work". No one got shiny golden handcuffs, but they did get their salaries paid for a year. Eventually someone from HQ figured this out and everyone got the boot.

I was the lone holdout. Instead of moving with the team, I'd gotten a nice offer from Adobe and decided to run with it. The good news was that my job lasted for more than a year. The bad news was that I actually had to work.


It's called "Vest in Peace" for a reason.

Your soul dies just a little bit (or a lot) while watching the clock.


rest and vest!


Getting paid for a pulse!


Sounds kinda hellish, honestly.


It certainly can be. However, if it's a 'golden handcuffs' situation a 'guaranteed' vest of several million dollars for a 3-5 year commitment is something that I would certainly be willing to take.


Is there a difference in the treatment/terms afforded a founder vs. those presented to a line employee at the acquired company?

Or is every case handled differently?


I think that's the point. Sometimes the acquiring company is hoping you'll get bored or fed up and walk out before you can vest. Why would they keep you productive which only serves to sustain your incentive to earn a couple million, when they can instead pay only a salary and hope you give up before your years are up?

I hate the definition of "acquihiring", which seems to have been written by a PR firm. It's made out to sound like an altruistic process that magically overrides the business's normal strategy where all that matters is the bottom line and bowing to investors / board members. The reality is that when acquihired, you've entered a new company with its own culture, underhanded and ruthless politics, and ulterior motives.

There is very little truly irreplaceable talent worth paying millions for. What an acquihire is really saying is "We need your help to transition this product into our company's structure. We can't do this in only 2 weeks, so we'll say now that you will be with us for years. Secretly we hope you leave shortly after the difficult transition is completed, but before we have to pay you".

If any business out there acquihires to the spirit of the definition, thanks for being of a rare species.


Write a letter to their customers that will get posted to https://ourincrediblejourney.tumblr.com shortly thereafter...


I spent a few years driving the integration of "our" technology inside Google. So yeah, there is agency.



Are you guys distinguishing between what the buyer pays and what the engineer gets?

To put it another way: if you personally promised the engineer $5 million (vesting over 3 years) they may switch.

This means the buyer must see some value in the company over and above the engineers. (IP? Customers?)


The engineers are not getting that obviously, it depends on how much equity they have and the difference between the amount of VC money pumped into a company and the final sale price of the company. VCs often get 2x preferred stock (at the minimum - it entitles them to 2x what they invested). This is before anyone else is allowed to see a dime. Every term sheet is different of course.


The value of the team is also larger than the individual engineers on it. Plus, a package bid is less risky if you're looking to get the entire package.

Oh, and poaching has another risk - that you get only the worst employees.


Depends on whether it is an all cash, all stock, or mixed deal.

Typically a "shotgun" clause is triggered and all employees immediately vest and cash out.

Founders, however, negotiate "earn out" clauses with the acquirer, which provides some upfront cash with the balance to be paid out over time, contingent on performance or retention goals.



I'm assuming Google doesn't need any of the company's business, and that it is being acquired for its talent.

I wouldn't assume that. Developer-friendly natural language APIs will only become more important over the next few years.


How's firebase going these days?

Edit: I should add Firebase could have(and should have) been the next PHP/Access. Could have solved almost every SMB business custom software need in a way that a business could go from front-end to front-end and never get locked into another 1,000 crud app for XYZ vertical. Could have wiped it all out. Firebase could have become the data store of the internet. Then Google showed up....


How's firebase going these days?

Given that Firebase seems to have become the foundation of Google's mobile app development offering I'm guessing ... pretty well?

¯\_(ツ)_/¯


Firebase uses a non-relational database correct? Shouldn't they then run into issues with problems that do not lend well to non-relational databases? I don't see how Firebase could have been anywhere near as pervasive as PHP.


My question is why would someone want to send their data to some api on the cloud rather than processing it locally himself using the numerous open source packages available to do this sort of thing?


For a lot of things, using a self-hosted open-source package is a good idea, absolutely.

However, for some applications there's a vendor with a neural network (or whatever) trained on a billion inputs, you see it perform better than the open source equivalent trained on a million inputs, and you don't have a billion inputs to improve the open source version with.


Congrats to them. I do find this sentence a bit awkward: "It has been a long and fun journey. Since API.AI’s launch in 2014..." I think the CEO referring to the total time into the company (including the Speaktoit phase), but it comes across badly - leaves me thinking "You think 2 years is a long time???" I don't think that's what the CEO meant to say.


My guess is it was an attempt to avoid saying 'incredible journey'.

https://ourincrediblejourney.tumblr.com/


What's the point of that page? Starting and selling your own company is an incredible journey. I don't see the problem with that phrase.


I'm sure it is an incredible journey.

It's just become cliché for companies to put up a page thanking customers (that are mostly likely getting screwed by the deal) for partaking in their incredible journey.

It's not so incredible when a service you were relying on stops providing that service.


Does this signal a continued push for bots and conversational UX?

Where does the belief in the tech come from? Are there any examples of real success with this tech?

Bots can't solve discovery in a novel way without big advances in NLP. There doesn't seem to be a bridge from here to there where it makes sense as a user.

Is this just irrational exuberance?

I'd love to understand why microsoft/google/facebook are all continuing to bet big here.


> I'd love to understand why microsoft/google/facebook are all continuing to bet big here.

Because they can afford to. And because if you wait for success examples, you might just be too late.


The bridge is a hybrid model where you can have humans responding and an AI in the back learning. With enough training data the hope is that the AI would take over and be able to do the tasks.

That being said, I actually haven't seen a successful implementation of a full transition from a hybrid to a chatbot example (maybe like x.ai or something, but nothing on a large scale). Think there is really good opportunity there for some basic tasks, but still see bots as glorified menu options for the time being.

Having worked at 2 of the firms you listed above and doing some deep V1 bot explorations for one of them, the immediate idea for bots in the short-term is to replace apps (which is not the best UI interaction model for tail-end apps)


Do you mean discovery as in me figuring out which bots would I like to install from a "bots store"?

If so, I think Apple's approach in iMessage widgets is fantastic: If you send third-party content to a person that doesn't have the plugin installed, they'll also get a link to install it. So it boosts word-of-mouth.


I could have been clearer because you're right that it's ambiguous.

My sense of discovery was meant in the UI sort of respect. When a feature is hidden from the user, that feature might as well not exist.

Bots (and command lines, for that matter) are awful unless you know exactly what they're capable of. There is an extrinsic knowledge you need.

That is, you can't expect a bot to know whatever you throw at it but you're unsure what you /can/ expect it to know.

The problem is that command lines are sort of made for the expert users whereas bots are said to be the opposite.


The beauty of (NLP) bots is supposedly that you can just ask what they can do.

Whether that's a viable UX strategy is still up for debate.


Ah, I agree. Bots need to understand language (within their domain) very broadly to be able and overcome that.


I know of a few large scale projects successfully implementing conversational UX with huge economies of scale. Most of the customers haven't even realised they are talking to bots.


Can you name a few (besides like Operator or Magic, which I don't even know would qualify since they're more hybrid)? I've done some explorations into bots for a project I worked on, and most of it just seems to be glorified menus.


Congratulations to the api.ai team. Unrelated to the acquisition directly, but fantastic logo placement. I can't be the only person that looks for their color-changing sign every time I drive down the 101. Awesome advertising / growth hack. :)


I think a sign on the 101 might be better for advertising to acquirers than to end users.


+1. Success factor, API.ai, Evernote — three logos from my memory


Yeah their logo is one block away from my apartment so I saw that 7 days a week


Motorola has one over by Moffett Towers if you are looking for a new muse :)


Don't forget Rambus behind them!


Where in 101?


In Redwood City if i recall correctly.


Tried API.ai a few months ago and was very unimpressed with the results (things were slow, expensive and not very powerful. Rebuilding the models was also broken at times). I am unsure why Google are acquiring them, especially given that they have DeepMind and TensorFlow teams.


Can it do anything that can't be done locally using off-the-shelves packages like nltk etc?


This kind of text reads like something that is going to end up in https://ourincrediblejourney.tumblr.com/


When I read this kind of stuff, I feel like it never makes sense to sell out if you care about your company unless you wanted to get rich all along. I just feel like the majority of companies that get bought up die a slow and painful death - and sometimes a quicker but equally painful one. It would pain me so, so much to see all the work I put into something like that go down the drain.


It would probably pain me too - until I looked at my bank balance.


I'm pretty sure the second line is a tongue-in-cheek reference to this meme. At least I'm hoping it is.


So what? Don't you think starting and selling your own company is an incredible journey?


I wonder how engineers at Google feel about all this. I guess they feel they should be joining these startups instead of slogging at Google. They can just easily retire/do their own thing if they are part of these acquihires.


Working at a startup "for the money" is like playing the lottery, you may hit it big, but a lot of startups just crash and burn.

In the meantime you're making a lot less than someone working at a large company. You could take the $50K/yr pay difference and buy lottery tickets with that.


You are right but it's not like the startup is not paying you anything. So it's not exactly the same odds of a lottery. Most startups pay from 100-125k and Google is not that far off from that range.


The gap in base-pay+bonus is usually like 20-30%, plus there's the fact that you're getting stock you can actually sell on top of that.


With Firebase expanding to form an entire ecosystem of services, and being kept alive for a long time now, I have renewed hope that being acquihired by Google won't just be the death knell - at least for this type of service.


Was Firebase acquihired? I would think not. Without knowing the details of the amounts paid, the fact itself that Firebase is still around and Google seem to actually keep investing in it - tells me it was more than just an expensive hiring process.


That's interesting they just bought apigee which does about the same thing?


Consolidation.


Facebook acquired wit.ai in January 2015, and I wonder if this acquisition is Google's quite belated response to that?


Could be, if you think about it. "Api.ai" is very similar because it also has a three letter name and an ".ai" /s


Another one of those services I wish I'd heard about before its acquisition. But at least this one does not appear to have been shut down on day one?


I'm actually excited for this. API.ai us a great set of tools to use.


I honestly don't think this is a aquire-hire. It seems much more of an attempt to get access to additional data not made public from companies using api.ai, at the same time boosting Alphabets capabilities.


Of course they are, what else would they do; ~2 years old and already sold [1]

[1] https://en.wikipedia.org/wiki/Speaktoit#Api.ai


Fucking impressive.


not sure what google gains by doing this, hope google can leverage this and give its customer more AI-APIs, the more the merrier, especially better speech-APIs(similar to Alexa but more open)


Literally never heard of this company. Serious question: How can I find a company that is likely to get acquired by Google?


Companies that solve a problem Google is facing (i.e. look at Uber acquiring Otto).


They seem to buy a lot of useless stuff though.


Congrats to the api.ai team.

I love when great companies do well.

I'm a happy customer (and wish I were also an investor, but I'm not).


> I'm a happy customer

What are your plans for transitioning off their service?

The script for this sort of acqui-hire is pretty well known at this point.


That's true, though I suppose Waze is still working, and Nest was allowed to continue operating too.


n alive = 2

n dead = too many to count


Google acquired APIgee and now API.AI. Why Google is so interested in API's now?


True they both have API in the name, but api.ai is more of an API for AI tasks like intention inference and sentence parsing. Apigee is an API management platform.


What they did? Yet another interface wrapper to Stanford Core NLP for over hyped chat bots?


> It has been a long and fun journey.

Using the word "journey" is this context is rather... risky.

https://ourincrediblejourney.tumblr.com


TBH it has got to the point now where I just assume people writing these posts are in on the joke and doing it on purpose. At least, I hope that's whats happening.


Why? What's the problem with calling it what it is? That site smells jealousy.


Why are you so intent on telling everyone off for criticising what, at the very least, is a unimaginative, cliched, and clumsy way of describing a cash grab.


Because it's ridiculous and I'm tired of seeing that site in every acquisition thread.


I guess everyone else is tired of seeing the same old rhetoric every time a company sells — it's great news for the founders and maybe the employees, usually not so great news for the users.

There's a podcast called "The Distance"[0] that investigates businesses that have been serving their customers for decades. Those stories are "incredible journeys". Two years of development and marketing ending in an aqui-hire isn't an incredible journey, and it certainly isn't a long journey.

Selling to Google is an accomplishment. It's an interesting accomplishment, but the rhetoric of journeys and battles is tedious. These people have an idea, implement and market it well, work hard, and achieve a solid exit.

They haven't changed the world, done anything particularly praiseworthy, or struggled very hard compared to the "journeys" of millions of other people. They built a business based on a promise to their users, and then more or less said: "screw you guys, we're off to Google now. Thanks for the journey." Pompous SV bullshit.

[0]:https://thedistance.com/


Why, it's usually indeed a journey, with a liquidity event as a destination. Often the the acquisition also means dissolution, so the destination is final.


And where does this leave the customers? Simply faceless, nameless trials that the hero-startup must conquer on its way to liquidity?


I mean, it's kind of a risk you face when you buy a service from a startup, whether they tell you or not.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: