Palantir's Amazon Competitor...Is Having Issues

+ a deeper look into why AIP is transformative

It’s Not That Easy to Compete with Palantir

A few weeks ago, Amazon CEO of AWS unveiled one of the secret projects they had been working on: an internal chatbot businesses could use throughout the enterprise.

Initially, many — including myself — saw this as some level of competition for Palantir. The platform was called Amazon Q:

I had two thoughts:

1/ It will be some type of competition to Palantir’s AIP — only because Amazon will not try to make money off of it but rather offer it as an additonal service to AWS customers, and given their distribution, will be able to charge significantly less than Palantir in order to get clients to buy more compute

2/ While it may be competition, it will further the market for AI enabled platforms in the enterprise and introduce millions of customers to what a platform like this could look like — if the platform wouldn’t be top-class, clients, now that they know platforms like this exist, could enquire about other companies like Palantir to procure real generative ai applications…

And as of yesterday, Amazon Q is having some issues…

I don’t think Amazon Q will deal with these issues forever, but it points to how difficult it is to build real generative AI platforms that are actually usable in the enterprise. Palantir’s focus on Ontology over the past 2 decades has allowed them to build the one product that decreases hallucinations and maintains governance and security of the data.

More big companies fighting for this space means the pie is getting bigger, and Palantir just has to continue being a strong horse in the race.

Speaking of horses…

HISA and Palantir Develop AI-Enabled Tool to Help Identify Horses at Increased Risk of Injury Before They Race

From the latest PR you may have missed because it wasn’t on Palantir’s website:

My takeaway from this partnership: Palantir, again, finds a way to work with so many different clients…

I didn’t even know HISA was a thing. However, every company in the world is generating massive amounts of data, and Palantir is up there when companies are thinking of how to build enterprise platforms to effectively manage that data and optimize their workflows via AI.

If the Horseracing Integrity and Safety Authority is using AI…the entire world will be using AI. This is Palantir’s market to capture.

Where Should Palantir be deploying their cash?

This was a great chart and question posed by Emir yesterday — most of the money Palantir is generating in cash is going into short term treasuires to lock in interest income which is helping them show net income and GAAP profits — I don’t think this is going to change for Q4, but could you see them allocating that capital anywhere else?

From Emir:

Palantir’s Ontology Augmented Generation (OAG)

Retrieval Augmented Generation is a process LLMs use in order to pull data from a vector database that has collected the data and gone through a process of embedding. As a result, the LLM can find accurate responses and supercharge productivity for someone looking for value from it.

Palantir’s AIP is not just using RAG, but rather Ontology Augmented Generation — leveraging the Ontology Palantir has built via Foundry and pulling from that ontology’s information, synthesized via an LLM (or K-LLMs as Palantir showcased at AIPCon, multiple LLMs working together to provide the best answer) in order to add more context to the reality of the organization. Here is the video where they show how AIP is doing this in an example about trying to find surgical masks and connecting via various suppliers.

THIS IS A BIG DEAL.

Above we talked about Amazon’s Q already hallucinating and leaking confidential data - if the platform can’t be trusted in the enterprise AND doesn’t have a solid ontology to ground the model in giving correct answers (the ontology), then it is essentially useless.

Now, imagine being a client and Palantir showing you an example like they showed above on your own data. The time to value for a customer to see how powerful this can be hopefully can convert to stronger customer growth at scale.

Also…

In the video that Chad, a FDE at Palantir released, he used my name as an easter egg to be the name of the AI co-pilot generating the answers.

This was the coolest thing in the world, thank you Chad and Palantir for engaging with the community <3

Closing Thoughts

A couple of people ended up subscribing to the paid version of the newsletter yesterday, which is funny to me because I didn’t even link a button to subscribe to the paid version so I don’t know how you guys found a way to subscribe since Substack makes it very complicated HAHA. The “paid version” is just an $80/yr subscription - no extra content - just supports the newsletter.

It means the world to me. I know many people do want to support, so instead of subscribing to the newsletter, when I launch the tool that I am working on in January — that’s when you can show your support. It will be $5/month and an actual product that I genuinely believe every single person reading this will find valuable enough to pay for and not hesitate to pay for because of the value you are getting from it.

I truly appreciate anyone who finds value from the content enough to somehow go through the trouble to find the button to subscribe and support — I am not entitled to anyone’s attention or hard-earned dollars, I do this because I love building and naturally want to keep creating content in this space, so anyone who shows support has my gratitude beyond words.

That’s it for today - see you tomorrow!