Inside Palantir

My understanding is that Palantir is running an engineering mill&#x2F;sweatshop using fresh out of school CS grads (particularly from Stanford). That much of their time is spent simply writing client-specific code that imports and cleans up data from disparate enterprise systems. That its not challenging, innovating, or exciting.<p>Of course this is all anecdotal based on what I&#x27;ve heard from friends and a Palantir engineer I met at bar.<p>Anyone with more insight know if this is accurate?

Hi all -- I wrote the Palantir article. It&#x27;s really awesome to see all this discussion about it. As I say in the post, please don&#x27;t hesitate to contact me if you&#x27;d like to chat confidentially. I am always eager to hear any tips or new information. Find me on WhatsApp, Telegram, Signal, or encrypted email. Contact info here:<p><a href="https:&#x2F;&#x2F;www.buzzfeed.com&#x2F;williamalden&#x2F;inside-palantir-silicon-valleys-most-secretive-company?utm_term=.tmJy8Qexp#.iheRjVPw1" rel="nofollow">https:&#x2F;&#x2F;www.buzzfeed.com&#x2F;williamalden&#x2F;inside-palantir-silico...</a>

This article is much more interesting than its fairly predictable headline implies. While the secrecy of Palantir has served as clickbait for many years, there&#x27;s actually real news in this.<p>And the news is about the difficulties of scaling a services-heavy, on-prem software company that basically rents out forward-deployed engineers at 10s of millions of dollars per year. Especially when the software is an open-source stack, and the engineers are increasingly junior as the companies grows. That said, it&#x27;s still great at sales. Big numbers there...

I want to see what Palantir actually produces. Like does Coca-Cola get a monthly report that says &quot;Hey, you guys should bring back Cherry Cola in Montana for an expected 4% increase in sales&quot; or what? What does $1mil&#x2F;month actually <i>get you</i>?

FWIW, Gawker reporters have been making FOIA requests of various agencies about their contracts with Palantir...you can see the responses on MuckRock:<p><a href="https:&#x2F;&#x2F;www.muckrock.com&#x2F;search&#x2F;?q=palantir&amp;models=foia.foiarequest" rel="nofollow">https:&#x2F;&#x2F;www.muckrock.com&#x2F;search&#x2F;?q=palantir&amp;models=foia.foia...</a><p>Many of those requests came back empty, here&#x27;s one that produced responsive documents: the NYC Department of Finance -- $150,000 for a 6-month pilot of their &quot;Perpetual Server Core&quot; technology to do fraud detection:<p><a href="https:&#x2F;&#x2F;www.muckrock.com&#x2F;foi&#x2F;new-york-city-17&#x2F;palantir-contract-services-24354&#x2F;#comms" rel="nofollow">https:&#x2F;&#x2F;www.muckrock.com&#x2F;foi&#x2F;new-york-city-17&#x2F;palantir-contr...</a><p>edit: specified that the requesters appear to be Gawker reporters doing their own investigation

&quot;Those anxieties come amid a wave of staff departures. A chart from Palantir’s internal wiki said the departures through mid-April amounted to 5.8% of all staff, or an annualized rate of 20%. That compares to a departure rate of 13.6% in 2015, 12.2% in 2014, and 9.2% in 2013. Palantir paid annual bonuses in March....&quot;<p>Many companies have waves of departures in the spring. Bonuses are paid (as at Palantir), holidays are over, kids are about to finish school. A 5.8% departure rate in the beginning of the year cannot be &quot;annualized&quot; any more than fruitcake sales from December. And if it does end up that 20% of Palantir quits in 2016, that&#x27;d be totally normal attrition for a large company. Where I used to work it was 30% among software engineers, and even this was not a problem.<p>I imagine that a number of ex-employees go on to work in the industry whose data they analyzed at Palantir. This may help to explain why employees are willing to work for less than elsewhere. It&#x27;s because three years later they will work for much more elsewhere. Especially the hedge fund analysts.

Saw the aftermath of a Palantir project after the client parted ways with them. What a train wreck. There was a lot of hype around them but when one looked deep into what they actually did it was a lot of smoke and mirrors and not a lot of substance. Even basic stuff like data cleaning and integration was poorly executed. Senior leadership didn&#x27;t see value in what Panintir did, the project was cut off and they were asked to leave.<p>Not surprised in the slightest to hear they are struggling.

The problem with Palantir is the following:<p>Three letter agencies are paying and will continue to pay for Palantir consulting and products.<p>However Palantir has been unable to fully productize their solution (it is still pretty much consulting). Thus they have hard time convincing Fortune 500 companies to pay due to the costs and depending too much on human interaction.

I don&#x27;t think this is all that surprising. If you look at their staffing numbers and their fundraises, they appear to not be sustaining business but just growing staff because that&#x27;s what startups do.<p>&gt; Lisa Gordon, said that “the majority of the company’s customer relationships are multiple years in length, and many are as long as 10 years.”<p>This is not good at all. Customer loyalty is important, but considering that they&#x27;ve been adding staff at a fast rate, <i>most</i> of their customers should be new-ish.<p>They&#x27;ve raised about $2.5b over an incredible number of rounds (multiple per year), and are currently bringing in about $420m.<p>Based on my understanding, they message as a software seller, but appear to make most of that revenue off of consulting and integration services tied around their software lock-in. They also message as a big-data company, but AFAIK don&#x27;t provide anything that would be called &quot;big data solutions&quot; these days.<p>edit: I thought Sankar&#x27;s name sounded familiar, turns out he was the guy at the very heart of Palantir&#x27;s very embarrassing industrial espionage and racketeering efforts against a competitor. He was apparently punished by being promoted to company president.<p><a href="http:&#x2F;&#x2F;venturebeat.com&#x2F;2011&#x2F;02&#x2F;16&#x2F;palantirs-third-black-eye-i2-lawsuit-settled&#x2F;" rel="nofollow">http:&#x2F;&#x2F;venturebeat.com&#x2F;2011&#x2F;02&#x2F;16&#x2F;palantirs-third-black-eye-...</a>

I don&#x27;t see this as particularly damning to Palantir. From what I understand, Palantir is really like a McKinsey&#x2F;BCG&#x2F;Bain that specializes in data rich projects. If you view Palantir as another consulting firm, I&#x27;d be curious as to how its rates and deliverables compare to the that of MBB. My uneducated guess is that what Palantir is offering for its billing rate is probably in line with the standard for consulting firms.

That picture of Alex Karp is amazing.<p>If I ever get too rich to never ever give a f*ck, I will dress like that.<p><a href="https:&#x2F;&#x2F;img.buzzfeed.com&#x2F;buzzfeed-static&#x2F;static&#x2F;2016-05&#x2F;6&#x2F;12&#x2F;enhanced&#x2F;webdr05&#x2F;longform-678-1462551369-3.jpg?no-auto" rel="nofollow">https:&#x2F;&#x2F;img.buzzfeed.com&#x2F;buzzfeed-static&#x2F;static&#x2F;2016-05&#x2F;6&#x2F;12...</a>

The article hints at a larger problem that underlies all data-science-as-a-service outfits: pricing for profitability. How can you price the generation of insights so that you; Palantir etc. can become profitable, while those insights can save on costs for the respective clients? Coke would need to generate $18M+ per year from the insights alone to justify the costs.

Confusing company on many levels. On one hand such huge clients, top talent (until recently), and Thiel&#x27;s famous success pre-requisite of having unique offering. But then ... they go and do really, really weird stuff like making recruitment videos shot on someone&#x27;s Iphone: <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=PhMqPoCQ5Q8" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=PhMqPoCQ5Q8</a><p>I think for a long time they got away with seeming really cloudy and potentially mis-managed because of the secrecy that is inherent to what they do. But clients are now asking tough questions about what value they can actually bring to the table for crazy dollars, and things may becoming home to roost.

Spinning the reported facts another way, Palantir has doubled revs over last year, and could turn a profit at will. Meanwhile it is able to set price optimally at verge of pain point for some of the largest enterprise customers. Sounds like a well-run company.

&gt; “One of the things we did well early on was to recognize and invest in the unique talents of each Palantirian”<p>I find it a bit weird how tech companies make up names for their employees now. Googler, Palantirian... does this make people work harder because they feel like they belong or something?

There seems to be a lot of confusion here so let me clear up some things from my former several years at Palantir.<p>1) Palantir&#x27;s product is a legitimate product that works quite well for many graph analysis use cases. If you can model your problem as a graph, chances are that Palantir will help you find some solid insights (if there are insights to be found) across your 15 disparate datasets. Software Engineers build the core platform.<p>2) Forward Deployed Engineers customize and extend the platform for specific client use cases and get client data into Palantir. If these customizations end up being useful, they get rolled into the core platform(ideally).<p>3) Palantir works their engineers very hard and the performance bar is <i>very</i> high. Calling it an engineering mill&#x2F;sweatshop is a bit much considering you get 3 meals&#x2F;day, massages, chiropractic, nap rooms, excellent events, laundry, etc and 6 figure salaries or massive stock or both.<p>4) Lots of people have been leaving. Part of this is because lots of oldies are hitting their 10 year marks and part of it is because Palantir went through a massive growth spurt over the last 4 years or so and now the fat is being trimmed&#x2F;people are leaving after concluding that it wasn&#x27;t a good fit&#x2F;they don&#x27;t want to work that much. If you can get into Palantir, you can get into Google: Why not go to Google?<p>5) If you want to see the actual product, just look around. Here are like 150 examples: <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;playlist?list=PLCA98B156F7EFD6A0" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;playlist?list=PLCA98B156F7EFD6A0</a><p>6) The bookings vs revenue thing is pretty deceptive but everyone knows that the money isn&#x27;t real until its in your bank account or there&#x27;s a signed contract for it to be in your account.

At least they have the decency to tell you who they work for right in their name.

Palantir is probably worth less than the money invested. At $420 million in revenue, they would need to be worth more than 6x revenue to match the $2.5 billion invested. They certainly have &quot;special sauce&quot; over and above their value as a consulting firm (which are worth maybe 2x revenue).<p>But how big is the market for that &quot;special sauce&quot;?<p>The intelligence customers are low-capability, have massive datasets and really do need Palantir. But they&#x27;ve tapped out that market. So look at the consumer brands in their customer list: low capability, yes. But you could fit the data for any of them into the RAM on one server.<p>Yes Palantir has super smart guys who can find fascinating relationships in the data. But there are only so many relationhips to find. And once that&#x27;s done, the IT staff of the customer can do the work easily themselves.<p>So I&#x27;m sorry but I think Palantir is a washout (worth less than the money invested). Nice offices though.

I&#x27;m actually surprised that Palatir paid below market rates. I heard they paid well above market rates, which justified their stance of never going public.

Whatever you do, please don&#x27;t discontinue Plottable.js

$1 million&#x2F;month. I think that would make even patio11 blush. :)

20% annualized turnover does not seem alarming for a company in the valley that hires mostly fresh college grads.<p>That implies an average length of tenure of 5 years. Very few college grads that I know stay at their first job for 5 years.

&quot;The company, based in Palo Alto, California, is essentially a hybrid software and consulting firm, placing what it calls “forward deployed engineers” on-site at client offices.&quot;<p>This says everything about Palantir, and depending on your perspective, their valuation is (not) justified.<p>It&#x27;s justified if you think of the equation Palantir = McKinsey + IBM. On the other hand, no consulting firm should have 20x revenue multiple.

How can they ask for such high prices?

To get an idea of the kind of stuff Palantir do then it&#x27;s worth taking a look at the presentation that was leaked during the HBGary leaks: <a href="http:&#x2F;&#x2F;www.businessinsider.com&#x2F;palantir-wikileaks-2011-2?IR=T#-1" rel="nofollow">http:&#x2F;&#x2F;www.businessinsider.com&#x2F;palantir-wikileaks-2011-2?IR=...</a>

I used Palantir heavily in my previous work and it was disappointing at how manual it was based on what we were sold as a mostly automated platform.<p>It had a great interface for creating network diagrams, and collaborating but it was a huge pain to get integrated.<p>I mean if nothing else, give us keyword matching and linking!

Interviewed with them once: took like 2 months and it was a horrible experience. Weirdest thing was I randomly ran into this guy onsite that was a friend-of-a-friend.. and enough of an acquaintance that we had each others&#x27; numbers and went drinking a couple times... saw him there wearing the same Palantir t-shirt as everyone else (which was weird) and he barely acknowledged me... like I wasn&#x27;t &quot;one of them&quot; yet... it was creepy enough that I was glad I didn&#x27;t get an offer.

FYI:<p>What does Joe Lonsdale think of BuzzFeed&#x27;s Inside Palantir article?<p><a href="https:&#x2F;&#x2F;www.quora.com&#x2F;What-does-Joe-Lonsdale-think-of-BuzzFeeds-Inside-Palantir-article&#x2F;answer&#x2F;Joe-Lonsdale?__snids__=1650809443&amp;__nsrc__=2" rel="nofollow">https:&#x2F;&#x2F;www.quora.com&#x2F;What-does-Joe-Lonsdale-think-of-BuzzFe...</a>

The headline graphic really caught my attention, because my fingers have ls -latr ingrained into them.<p>I thought it oddly fitting.

I know a few people who have worked for Palantir. They all agree it doesn&#x27;t work. They all quit because Palantir data analysis is being used to ruin lives (throw people in jail) and they believe the whole operation to be a giant defrauding of the government.

Apparently they&#x27;re not doing well. Good riddance. Palantir is creepy as hell. For all the snark about how startups and places like Apple are run like cults, like in The Circle, Palantir is the one outfit I know of where the stereotype is really true. And in the post-Snowden world where we know for sure just how all-devouring the US intelligence&#x27;s community&#x27;s desires are regarding our privacy, the last thing we need is Silicon Vally outfits helping them out. Assuming Karp can&#x27;t right the ship again, I will not be sorry to see them go.

&gt; “We’re looking to do transformational work with our customers,” Gavin Hood, Palantir’s chief of staff, said in an interview with BuzzFeed News. “Finding the right partner to do that transformational work takes a lot of care and a lot of attention.” He added, “There’s a lot of reasons why that doesn’t always work out.”<p>Sure it&#x27;s hard to find suckers who will give you million of dollars for poorly defined return.

Theranos, Zenefits, palantir, what&#x27;s the next unicorn deflation? Is it Uber or Spotify with their Palantir-esque spend a $1&#x2F;make $0.90 strategies or is it an another company with less-reported vulnerabilities?

&quot;Although the company is not profitable&quot;... Excuse me but, you have to be fucking kidding me!!

Will Alden is shaping up to be one of the most connected journalists in startupland. First the Zenefits coverage, and now this, he&#x27;s done a great job getting lots of great sources at some of the more interesting companies.<p>Credit to BuzzFeed for building up a pretty solid news team.

&gt; If you have information or tips, you can contact this reporter over an encrypted chat service such as Telegram, Signal, or WhatsApp, at 310-617-1302. You can also send an encrypted email to will.alden@buzzfeed.com, using the PGP key found here.<p>This is the best &quot;contact me if you have juicy insider info&quot; slug that I have <i>ever seen</i>. He can take PGP-encrypted email!<p>And this is on Buzzfeed!!!

This article was way, way better sourced than I thought it&#x27;d be. Especially for Buzzfeed of all places.

From a linked article:<p><i>&quot;The primary payday for the best engineers is that you get to work with the best engineers,&quot;</i><p>Funny how that&#x27;s not the same for execs!

There are a lot of similarities to Theranos.

SAP-like scam. Highly refined, Oxford-educated sales and execs, selling with high-status deceptive techniques grossly overpriced, lovest quality outdated by a decade crap, for support and maintenance of which they are billing their victims for each hour spend by expensive suit wearing nonsense talking consultants.<p>BTW, the guys who went there for sweatshop positions are those who barely managed to graduate. But in sales there are top tier liberal arts guys, with refined speech and behavior.<p>These kind of enterprises is a classic social pyramid, modeled after an organized religion, especially Catholic Church, with exactly the same deceptions, dogmas and loyalty for lower ranks.

I&#x27;m curious if any positions require pre employment drug tests, random urinalysis, or security clearances. What a miserable life for a 20 something in the private sector!