Man, you are the best explaining what palantir does. It is something different than the other youtubers with their insanely price targets... Like From me.
Hi CodeStrap, a truly excellent and articulate presentation of PLTR. Watching this now in Sept 2021, it seems that many of your foresight thoughts and ideas you put forward in April are indeed coming to fruition in PLTR. Expanding into smaller markets by investing into sales. Investing in SPACS to span across many different verticals within highly innovative companies in their respective spaces (most seem likely to become leaders within their spaces). To list but two initiatives. Many thanks for making your videos. I find them so insightful. Regards Howard
Hi Codestrap, Thanks for sharing your insights into their software. I appreciate these since I’m not versed in data science or computer science at all, and your explanations help me understand palantirs possible applications.
Good points. Palantir is definitely 10 times more advanced than its "competition". Just looking at its platform are all the business solutions that others can't compete. Palantir has a few more years creating technology(platform+services) that doesn't really exist, making it a leader in everything that comes its way.
Thanks for the great video! just subbed. I also think Palantir has amazing products (esp Foundry). I'm personally hoping to see it trump all data-related SaaS out there and solve "SaaS fatigue" That said, I have some points I wanted to ask you. First one is regarding their filtering part of life sciences demo in double click. I actually didn't find that impressive at all, filtering is a very common feature in any enterprise software. Sure it's more dynamic and not hardcoded to a single column/field but I'm not sure if that was worth filling up the already-short double click demo, unless I'm missing something? Most importantly though, my biggest concern is "product bloat". What I mean by that is how many enterprise tech companies has a huge directory of products (often duplicated logic but just different context) like how SAP, ServiceNow has 20+ redundant products in their product directory. There are so many companies with overlapping products as well, I feel like everyone has some form of "log monitoring" product, which makes me feel like "product bloat" is a serious problem for the user's end. Do you think Palantir Foundry will ever suffer this problem? It seems like they are trying to work with tens of industries that I feel like they have been and/or will need to build more features/archetypes only applicable to certain industries, leading to huge "product bloat", making Foundry very hard to use (or could I be wrong in that it won't necessarily be hard to use?).
Totally agree on the first point. There is a technical term for the second one. It has to do with product lifecycle, but the word escapes me. Yes it's a concern, but it's a concern for everyone and has been abstracted to some best practice. The best solution to this problem IMO is always subjecting product requests to a YAGNI test. I wrote an article about this on Medium called "Cracks in the Walled Garden". It dives into issues I see with AWS related to this very problem. It all comes down to the quality of the folks running your company really. PLTR has a top notch team so I don't think they will struggle with this anymore than anyone else has.
Palantir also provides customizable modules like table or dossier or AVA. Google “Palantir modules”. Although the products may be drastically different in various industries, the underlying modules can be customizable and reusable. When the modules are organized based on functionality, i think it is a matter of days or weeks to piece them together as a application or prototype.
hey CodeStrap, great video here and a lot of industry insights. Speaking of ML/AI, I actually don't recall any in depth mention of how they empower AI models or how AI is used to drive decision making in their demos. Did I miss something?
Let's be clear on what AI actually is first. AI is robotics + machine learning + deep learning. I would actually assert that in practice it's robotics + computer vision + NLP (we are nowhere near general AI). PLTR empowers organizations to pursue AI through platform tools that enable ML intiatives. PLTR's archetypes appear to be industry-specific solutions that include ML that is ready to apply to a category of problems. I think we are both missing something though, hands-on experience with the platform. Lot's of questions and grey areas here. So whatever you do, do your own research.
Thanks for the great video! Any thoughts on the advantages Foundry has over Alteryx? There’s some videos on CZcams covering Alteryx that looks similar to Foundry. Was wondering your thoughts from a technical perspective?
Hey there, thanks for the comment. It looks similar. I am skeptical that software engineers who are not data scientists, or at least supported by data scientists, can deliver business value. That seems to be the biggest problem with AutoML solutions and platforms today. You just can't deliver business value without both the tools and the experts. Like I said in the video, there are open source tools you can use right now to enter the market with a similar platform. But at the end of the day these tools only get you part of the way. You need experts to get you over the finish line. Making this process viable is a super hard problem though because you need to sell both a service and a platform. So what makes PLTR unique is the 10 year hard start in figuring out how to do this. The question in my mind is who will be first to get viable ML/AI tools to market that are 100% complete. I still think that is a short list of companies which PLTR is on. Why? Because its all about the process of iteration and feedback loops to make this breakthrough. PLTR is way out in front of establishing these loops and has been iterating on the problem way before even Amazon knew this was the future of software.
Thanks for the video! What are your thoughts on (seemingly) less complex issues such as data ownership and maintaining master (single source of truth) data? I see this as a more direct reason for most large companies to want a platform like Palantir offers. Example: An auto manufacturer like BMW may for example have a design & engineering team developing a new platform - or car model if you will - where that team have ownership of the bill of material for said model. The aftersales team that owns the maintenance business and are responsible for setting maintenance recommendations at specific mileages may run predictive analysis using a combination of the bill of material and maintenance data (e.g. what spare part typically breaks at what mileage). Often times the bill of material for the model is duplicated by the after sales division and the dependency is lost to their design and engineering team. Meaning that, if the bill of material is changed - then the maintenance data becomes inaccurate and invalid - to some degree. In reality there may be hundreds or even thousand of dependencies derived from the bill of material to various tools - like CPQ, supply chain analysis, sales analysis etc. Tracking the ownership of data and ensuring robustness is extremely difficult - especially considering that a lot of the data may come from suppliers (think BMWs complex supply chain) and be outside the control of the company. This approach where new silos containing duplicate data are created is very common, and its really difficult to do it any other way. Creating robust dependencies requires IT-projects that takes a lot of work and a long time to implement - slowing the business down. Palantir seems to solve this problem through it's "purpose" driven data ownership functionality working somewhat like git works for software development. Not a very sexy problem to solve - but it creates a lot of business value and is the first step for any business to reach a digital maturity where AI/ML is enabled. It is moreover - in my opinion - also the most difficult digital challenge of any non-tech business. The way non-tech firms are managing their data is equivalent to tech-firms doing software development without git.
I think there are actually several problems to unpack in your example. So I'm not confident in solving the hypothetical scenario without a lot more information on the actors, workflows, etc. But generally you are describing the exact problems blockchain is meant to solve. Immutable smart contract that can enable trustless (or semi trustless) transactions with immutable audit trails is a pretty big hammer in the supply chain space. Not having used Foundry I can't speak to its ability to mitigate these problems. The type of access controls they demoed generally allow fine grain data access for research teams. AWS Glue Data Catalog has similar features. They essentially allow you to define multiple views of data with fine grained column level access that can be shared across teams or even organizations. Without developer documentation and some hands on experience though it's really hard to know how I could apply it to the hypothetical scenario you described.
@@codestrap8031 I am really looking forward to the day you get first-hand access to Foundry. Do you think that day will come anytime soon? Much love from Sweden
Hi -awesome videos. I watched palantir DC couple of times. What exactly is archetypes ? Are they solutions built on top of palantir platform or is it much deeper than that ?
Yes, they are pre-built modules that include data models and automation for customers in specific markets that are part of PLTRs platform. A lot of these companies use the same set of ERP suites and software solutions. Once you deconstruct one of them into a modern ML/AI stack you can essentially deconstruct all of them. Without actually getting the platform in my hands though that's about all I can say with confidence.
Can't link the article as I don't think CZcams will allow it in the comments section. Just head over to medium and do a search for "Operationalizing Data Science". The book is called "The Architecture of Privacy" from O'Reilly.
It's a joke. I did not mean for it to me a firm price target. I just meant to underscore the fact it isn't going to experience explosive growth until they make the platform more turn key. There's an inflection point somewhere out there in the next 3-4 years IMO. But until then I think they are going to work hard to justify that 50 billion(ish) valuation.
@@ASUKiller Next few months should go sideways due to deleveraging, increased margin requirements on stocks like PLTR, msm fud; however, with a couple strong quarters institutions will begin to build strong positions and EOY I see it around $35, in 5 years $150-200 so 10x investment potential, downside is limited, so under 22.5 now for shares would be very fair risk/reward, but stay away from options this is a buy and hold shares for years type of investment.
Man, you are the best explaining what palantir does. It is something different than the other youtubers with their insanely price targets... Like From me.
Yes, totally agree.
About to watch, but I'm so excited you released a video again. Your insights are absolutely terrific for more technical people
Hi CodeStrap, a truly excellent and articulate presentation of PLTR. Watching this now in Sept 2021, it seems that many of your foresight thoughts and ideas you put forward in April are indeed coming to fruition in PLTR. Expanding into smaller markets by investing into sales. Investing in SPACS to span across many different verticals within highly innovative companies in their respective spaces (most seem likely to become leaders within their spaces). To list but two initiatives. Many thanks for making your videos. I find them so insightful. Regards Howard
Thanks for the analysis. I have been waiting for this video for a while.
Thanks! Very valuable information! Keep up the good work! 🙏
Thanks for posting. Your vids make understanding palantir much easier for people like me who don’t live in the data science world.
Great video! Thank you for the unique insight.
Thanks for another great video helping me understanding the value proposition.
Hi Codestrap,
Thanks for sharing your insights into their software. I appreciate these since I’m not versed in data science or computer science at all, and your explanations help me understand palantirs possible applications.
thanks a lot! very important to get analysis from the pro in the field!
This is just great insight to watch. Thank you so much for sharing this! 🙏🤩
thank you for the good work. so many information for a layman to learn
I love Palantir's value creation aspect
Simply, the best.
Good points. Palantir is definitely 10 times more advanced than its "competition". Just looking at its platform are all the business solutions that others can't compete. Palantir has a few more years creating technology(platform+services) that doesn't really exist, making it a leader in everything that comes its way.
Thank you.
Thanks for the great video! just subbed. I also think Palantir has amazing products (esp Foundry). I'm personally hoping to see it trump all data-related SaaS out there and solve "SaaS fatigue"
That said, I have some points I wanted to ask you. First one is regarding their filtering part of life sciences demo in double click. I actually didn't find that impressive at all, filtering is a very common feature in any enterprise software. Sure it's more dynamic and not hardcoded to a single column/field but I'm not sure if that was worth filling up the already-short double click demo, unless I'm missing something?
Most importantly though, my biggest concern is "product bloat". What I mean by that is how many enterprise tech companies has a huge directory of products (often duplicated logic but just different context) like how SAP, ServiceNow has 20+ redundant products in their product directory. There are so many companies with overlapping products as well, I feel like everyone has some form of "log monitoring" product, which makes me feel like "product bloat" is a serious problem for the user's end. Do you think Palantir Foundry will ever suffer this problem? It seems like they are trying to work with tens of industries that I feel like they have been and/or will need to build more features/archetypes only applicable to certain industries, leading to huge "product bloat", making Foundry very hard to use (or could I be wrong in that it won't necessarily be hard to use?).
Totally agree on the first point. There is a technical term for the second one. It has to do with product lifecycle, but the word escapes me. Yes it's a concern, but it's a concern for everyone and has been abstracted to some best practice. The best solution to this problem IMO is always subjecting product requests to a YAGNI test. I wrote an article about this on Medium called "Cracks in the Walled Garden". It dives into issues I see with AWS related to this very problem. It all comes down to the quality of the folks running your company really. PLTR has a top notch team so I don't think they will struggle with this anymore than anyone else has.
Palantir also provides customizable modules like table or dossier or AVA. Google “Palantir modules”. Although the products may be drastically different in various industries, the underlying modules can be customizable and reusable. When the modules are organized based on functionality, i think it is a matter of days or weeks to piece them together as a application or prototype.
hey CodeStrap, great video here and a lot of industry insights. Speaking of ML/AI, I actually don't recall any in depth mention of how they empower AI models or how AI is used to drive decision making in their demos. Did I miss something?
Let's be clear on what AI actually is first. AI is robotics + machine learning + deep learning. I would actually assert that in practice it's robotics + computer vision + NLP (we are nowhere near general AI). PLTR empowers organizations to pursue AI through platform tools that enable ML intiatives. PLTR's archetypes appear to be industry-specific solutions that include ML that is ready to apply to a category of problems. I think we are both missing something though, hands-on experience with the platform. Lot's of questions and grey areas here. So whatever you do, do your own research.
Thanks for the great video! Any thoughts on the advantages Foundry has over Alteryx? There’s some videos on CZcams covering Alteryx that looks similar to Foundry. Was wondering your thoughts from a technical perspective?
Hey there, thanks for the comment. It looks similar. I am skeptical that software engineers who are not data scientists, or at least supported by data scientists, can deliver business value. That seems to be the biggest problem with AutoML solutions and platforms today. You just can't deliver business value without both the tools and the experts. Like I said in the video, there are open source tools you can use right now to enter the market with a similar platform. But at the end of the day these tools only get you part of the way. You need experts to get you over the finish line. Making this process viable is a super hard problem though because you need to sell both a service and a platform. So what makes PLTR unique is the 10 year hard start in figuring out how to do this. The question in my mind is who will be first to get viable ML/AI tools to market that are 100% complete. I still think that is a short list of companies which PLTR is on. Why? Because its all about the process of iteration and feedback loops to make this breakthrough. PLTR is way out in front of establishing these loops and has been iterating on the problem way before even Amazon knew this was the future of software.
Thanks for the video!
What are your thoughts on (seemingly) less complex issues such as data ownership and maintaining master (single source of truth) data? I see this as a more direct reason for most large companies to want a platform like Palantir offers.
Example: An auto manufacturer like BMW may for example have a design & engineering team developing a new platform - or car model if you will - where that team have ownership of the bill of material for said model. The aftersales team that owns the maintenance business and are responsible for setting maintenance recommendations at specific mileages may run predictive analysis using a combination of the bill of material and maintenance data (e.g. what spare part typically breaks at what mileage). Often times the bill of material for the model is duplicated by the after sales division and the dependency is lost to their design and engineering team. Meaning that, if the bill of material is changed - then the maintenance data becomes inaccurate and invalid - to some degree.
In reality there may be hundreds or even thousand of dependencies derived from the bill of material to various tools - like CPQ, supply chain analysis, sales analysis etc. Tracking the ownership of data and ensuring robustness is extremely difficult - especially considering that a lot of the data may come from suppliers (think BMWs complex supply chain) and be outside the control of the company. This approach where new silos containing duplicate data are created is very common, and its really difficult to do it any other way. Creating robust dependencies requires IT-projects that takes a lot of work and a long time to implement - slowing the business down.
Palantir seems to solve this problem through it's "purpose" driven data ownership functionality working somewhat like git works for software development. Not a very sexy problem to solve - but it creates a lot of business value and is the first step for any business to reach a digital maturity where AI/ML is enabled. It is moreover - in my opinion - also the most difficult digital challenge of any non-tech business. The way non-tech firms are managing their data is equivalent to tech-firms doing software development without git.
I think there are actually several problems to unpack in your example. So I'm not confident in solving the hypothetical scenario without a lot more information on the actors, workflows, etc. But generally you are describing the exact problems blockchain is meant to solve. Immutable smart contract that can enable trustless (or semi trustless) transactions with immutable audit trails is a pretty big hammer in the supply chain space. Not having used Foundry I can't speak to its ability to mitigate these problems. The type of access controls they demoed generally allow fine grain data access for research teams. AWS Glue Data Catalog has similar features. They essentially allow you to define multiple views of data with fine grained column level access that can be shared across teams or even organizations. Without developer documentation and some hands on experience though it's really hard to know how I could apply it to the hypothetical scenario you described.
@@codestrap8031 I am really looking forward to the day you get first-hand access to Foundry. Do you think that day will come anytime soon? Much love from Sweden
Hi -awesome videos. I watched palantir DC couple of times. What exactly is archetypes ? Are they solutions built on top of palantir platform or is it much deeper than that ?
Yes, they are pre-built modules that include data models and automation for customers in specific markets that are part of PLTRs platform. A lot of these companies use the same set of ERP suites and software solutions. Once you deconstruct one of them into a modern ML/AI stack you can essentially deconstruct all of them. Without actually getting the platform in my hands though that's about all I can say with confidence.
Could you link the article you wrote on medium and kindly name the book you‘re reading written by the person working at pltr? Thank you!
Can't link the article as I don't think CZcams will allow it in the comments section. Just head over to medium and do a search for "Operationalizing Data Science". The book is called "The Architecture of Privacy" from O'Reilly.
Have you thought about working at Palantir?
No. I only work at companies that are pre-ipo.
@@codestrap8031 Why? Is that because you want to own pre-IPO shares of a company.
Or just the stress of working at public companies?
Palantir investors should go back to these videos so they can learn then maybe some of these weak investors will hold on to their stocks
$31 by 2025, I don't understand your valuation, please explain why only $31?
It's a joke. I did not mean for it to me a firm price target. I just meant to underscore the fact it isn't going to experience explosive growth until they make the platform more turn key. There's an inflection point somewhere out there in the next 3-4 years IMO. But until then I think they are going to work hard to justify that 50 billion(ish) valuation.
I am thinking to buy more, what price do u think is a fair price?
@@ASUKiller Next few months should go sideways due to deleveraging, increased margin requirements on stocks like PLTR, msm fud; however, with a couple strong quarters institutions will begin to build strong positions and EOY I see it around $35, in 5 years $150-200 so 10x investment potential, downside is limited, so under 22.5 now for shares would be very fair risk/reward, but stay away from options this is a buy and hold shares for years type of investment.
@@mr__christian , thanks a lot! I have a lot of options, I am thinking to close them, and switch to shares holding