🎸 The Woodstock of AI: Inside Palantir's AIPCon
AI is real and sounds loud.
Editor: Emanuele Marabella
Palantir aims to take the whole AI market.
Last week I saw this firsthand.
Based on the success of the previous editions, Palantir hosted the fourth AIPCon, which I had the pleasure to attend.
Let me navigate you through my experience at the Woodstock of AI!
Grab your Palancino.
A special thanks to Chad Wahlquist, ElianoYounes, and the entire Palantir team for making this possible!
The concert
We were at a corporate event, but the vibe was that of a concert.
The room was packed and the excitement was real.
Music on.
CEO Karp introduced the event underscoring how Large Language Models (“LLMs”) like ChatGPT are not enough to deliver value.
Without an ontology, AI models provide only “self-pleasuring and no fun.”
To deliver value you need to go beyond chatting.
“In the end you have products, your business, chips and ontology. This is how you get things to work.” - Palantir CEO Alex Karp
Palantir AIP is definitely not a LLM.
It’s a platform that helps customers solve use cases across any industry by governing the power of LLMs, while relying on the ontology as the “source of truth.”
LLMs are like genies of the lamp. AIP is the lamp that controls them.
Palantir AIP ensures genies perform their task accurately and safely.
Check my introductory article for more details:
The lights then turn on the real stars of the event: the clients.
The hit-song
8 keynote speakers from different industries explained how they are using Palantir to leverage their operations to solve real use cases with AI:
AARP;
SOMPO Japan;
NorthWind;
Tampa General Hospital;
Jacobs;
Fujitsu;
Parexel;
United Airlines.
I appreciated all the demos, but one in particular caught my attention.
Jacobs' EVP Shannon Miller presented how Palantir efficiently helps front-line engineers assess the health of water plans through Mixed Reality (“MR”) glasses.
Palantir AIP provides workers with visual AI recommendations and warnings by looking through the screen:
This is not a video game. It’s real software deployed into real operations.
Jacobs, which operates global critical infrastructure, represents one of the best Palantir customer journeys. The proof is their continued willingness to present their results with Palantir through the years:
The partnership makes both companies stronger as they continue their shared journey and build trust.
Another customer that already attended was Tampa General Hospital, which presented at FoundryCon on February 23 and announced a $50mn 7-year deal with Palantir just the day before AIPCon.
Deals speak louder than words!
After the keynote presentations, the live stream closed and we went to Palantir’s headquarters for the next phases of the event.
Meet the Masters
This section provided the chance to engage in informal discussions with a range of customer leaders in rapid-fire sessions of 10 minutes.
I joined the SwissRE one, hosted by Development Strategist Robbin Tops.
SwissRE has been working with Palantir since 2018 and is a key European client. According to a presentation shared in 2022, ~40% of SwissRE employees use Palantir Foundry.
I couldn’t lose the opportunity to learn more.
Robbin Tops explained that SwissRE uses AIP to leverage multi-modal workflows for three use cases. The main one is combining forensic police data with other data to assess the validity and severity of the claims.
In simple terms, the job of a reinsurance company like SwissRE is to understand the probability and impact of certain risks so that it can price accurately premiums to clients and diversify better. AIP helps SwissRE use LLMs to analyze unstructured data like videos and images to evaluate risks effectively.
Between the lines of the conversation, I spotted the key to Palantir’s success:
“You have experts that confirm the validity of the recommendations. With AIP you collect the feedback that helps the AI model improve.”
The ontology learns from the feedback of the users, facilitating the efficacy of further recommendations.
I promptly asked whether other companies provided similar solutions:
“I haven’t seen another application that can do that. Foundry is the only application that integrates all the processes and closes the loop.”
This is when I fully realized why Palantir is set to win in the AI space.
You can improve LLMs by giving them more data. However, there is an invaluable source of data that LLMs alone can’t capture: expert feedback.
To have increasingly more accurate recommendations from AI, you need to train it with the decisions and feedback coming from domain experts –that’s the most valuable source of information. Only an ontology allows you to “capture” that data so that it remains in your organization to empower following decisions.
The ontology allows knowledge to be compounded.
Only Palantir has been able to deploy this concept of ontology into complex organizations and its implications are stunning:
“It took me 2-3 weeks to understand to identify the problem I wanted to solve with the multi modal approach in AIP. Building the prototype was a matter of days.”
Not weeks or months. Days!
Without an ontology it would be impossible to:
capture feedback into the knowledge of your company (close the loop);
build new operational applications in days;
measure the real effectiveness of developed applications.
As the presentation ended a representative of an insurance company, a potential customer for SwissRE, approached Robbin Tops to learn more.
The network effects I described two years ago are real.
The AI hunt
The ~70 customers presenting at AIPCon were not paid, CEO Alex Karp said.
Why should they attend?
The audience was composed of a good mix of existing customers and prospective ones. I guess that ~40% were clients and ~60% prospective clients.
For existing Palantir clients, participating in AIPCon means:
advertise their AI efforts to their board and stakeholders. Executives need to show evidence of their “AI strategy”.
consolidate their relationship with Palantir. Palantir is proving its software work in the most adverse situations. Being close to Palantir becomes a way to link to its success.
For prospective clients, AIPCon allows:
build the relationship with Palantir at the cost of a day. Little to lose, and much to gain.
discover how Palantir could solve its critical problems by listening directly to customers who solved them.
assess a potential career opportunity. A successful implementation of Palantir associated with significant business output could provide a chance to rise within corporate rankings.
Participating there were also Palantir Partners like Taylor Gregoire-Wright, who previously worked at Palantir for 5 years and is now helping organizations become self-sufficient with Foundry & AIP with company Ontology. Partners like Ontologize are critical to helping Palantir scale.
I noticed a critical commonality among all participants: the desire to hunt.
A potential customer, Software Engineer Manager Christopher Knight, told me:
“I knew Palantir as a retail investor and being in data science. I am seeking an outcome-based use case to convince my organization to deploy to get a production use case in 3 months. Typically you don’t get it. Hopefully one day I will present too.”
Palantir AIP has already solved uncountable use cases across virtually all industries.
Therefore attendees of AIPCon are seeking:
Evidence that AIP can solve one of their problems by listening to direct experience from someone who solved them;
Inspiration from problems solved in different industries which could be adapted. In the same way, Palantir learns from transferring defense use cases to commercial ones and vice-versa, customers learn valuable intuitions on how to solve a problem by listening to the experiences of Palantir partners in other industries.
It’s hard to define the “standard attendee:”
I recognized a healthy mix between Fortune 500 and small companies. So far Palantir has been focused on the leading companies of each vertical, but it seems its focus is widening;
There was no dominant sector, but I was pleased to see many healthcare and heavy industries companies. Palantir shines amidst regulatory and complex operating constraints;
The audience included both C-levels and young executives.
AI for all tastes: Demo Expo
Customers showcased applications they built with Foundry & AIP across various industries and use cases.
AIP seems able to solve limitless use cases.
We are not talking about seeing a data dashboard. We are talking about the software used by front-line workers to deploy their daily work, in a fraction of the time, more effectively.
Dashboards allow users to see information. AIP makes users execute.
For instance, Trios Health uses AIP to schedule therapies efficiently:
By attending Palantir’s demo, I realized a “secret.”
Palantir uses AIP internally to improve its operations. Probably even Palantir is not using AIP at its full potential and by interacting with customers it can learn how to leverage it into use cases it didn’t think before.
For instance, Clare Kessi and Julia Richards showed how Palantir uses AIP internally to facilitate the screening of candidates’ applications by “digesting” the unstructured information from thousands of CVs and seeking relevant characteristics like “University” or “Past experiences.”
AIPCon is also an occasion for Palantir to showcase its beta products and receive feedback. For instance, Palantir is now releasing the AIP evaluator, which helps customers assess how different LLMs perform different tasks to choose which one or ones to deploy for each use case.
Within AIP, you can select the AI model you want at the cost of a click. With AIP Evaluator you can maximise LLM performance while reducing related costs.
Are AI models doomed to become a commodity?
Build the “Netflix engine” in minutes
The following phase of the event was called Bring Your Own Data (BYOD). This was a time dedicated to allowing customers to build a use case by leveraging AIP Now with their own data.
As I didn’t have “my data,” I attended the Advanced Bootcamp session with Jack Dobson, Commercial Lead.
In a few minutes and a few prompts, we were able to create a recommendation engine for books.
Stunning.
This is a notional example that can be the archetype for real complicated use cases for instance ticketing, customer care, supply chain, and promotion workflows.
While this could sound scary for data scientists, a system like AIP empowers them. They could focus on solving problems by building on the ontology rather than spending time trying to build tools that hopefully deliver results.
A demo like this would not be possible if the product was not exceptionally strong. This wasn’t a presentation, it was an application built on the go.
The new tool, AIP Evaluation, received particular interest from the participants.
“I’ve been working for a year on this!” the person next to me exclaimed.
All the people at the table were prospective customers who attended at least one BootCamp. They were seeking evidence Palantir can deliver and that is a trustful partner.
Will this experience convert them into clients?
Why does AIPCon succeed?
David Markey, Business Development at Palantir told me:
Palantir is not a tool, but a methodology to solve problems.
The methodology Palantir deploys is grounded on the belief that the closer you are to the problems the more you understand them. This way you can act backwards to solve them.
The environment Palantir can create at AIPCon facilitates a crucial exchange of feedback and ideas. People are open to discussing their issues, creating the condition for a potential solution.
Elias Müller-Möhl, Palantir Forward Deployed Engineer, explained to me that at these events:
“You learn what is possibile.”
AI is new and some software vendors are already exaggerating what their offering can achieve, essentially selling dreams.
In contrast, Palantir is so confident that by organizing an event like AIPCon it can show its software can solve use cases in a fraction of the time compared with other software vendors or software integrators.
Palantir is making the big bet that once customers become aware of what is possible, they will eventually turn to Palantir once they compare it with other possible solutions or, even better, after competitors fail to deliver their promises.
Palantir rocks
When you combine the most powerful AI toolkit with a curated selection of customers, the cocktail is explosive.
The general feeling is that if I were a potential customer I would love to at least try deploying AIP into a use case.
Proof sells better than any sales pitch.
AI may be the hot word of the moment, but the sentiment is real, and real results speak louder than words.
AIPCon: What’s next
Given the success of the event, I believe:
Palantir will continue to expand and improve the AIPCon and Bootcamp formats. My experience gave me a sense of a clear market fit between what customers search and what Palantir provides. These events are a great way for customers to become aware of a possible solution without passing through the usual PowerPoint slides and multiple steak dinners. A direct experience from an industry leader is worth more than 10,000 slides;
Palantir will gradually enlarge its focus on small-medium enterprises while showcasing industry leaders so that it can become the standard for related verticals;
Mixed Reality will become an increasingly important aspect of AIPCon presentations, especially for heavy engineering companies like Honda Heavy Industries, Airbus, and BP. I tried again the MR headset and a giant TITAN appeared in front of me. I could check the health of each single component, which made me realize how critical it is to have an ontology. Without it, it would be impossible to rely on the information I saw and therefore draw any decision.
The aftermath
The day went by extremely fast. I tried to live the experience as a potential customer rather than as a "Palantir creator” to have a clearer opinion.
This was the first event Palantir opened to a retail investor like me and Amit Kukreja, who brilliantly hosted the backstage interview with Palantir employees Chad Wahlquist, Jack Dobson, and Natalie Schneider during the break.
It seems Palantir is acknowledging that there is true value in the community to help spread the great achievements the company is achieving.
I am excited to share this journey.
My firsthand experience with AIP made me realize that if one day I build a firm, I would not think twice about building it on top of Palantir. The power of AIP is unbelievable despite the fact that product is only 1 year old. I see a future where entire new multibillion-dollar companies could be built on top of it.
Palantir suffered from a bad reputation for a long time under the “controversial spy black-box consulting company” label.
AIPCon showed me that not only is the label profoundly wrong, but also that the music has changed.
We are only at the first song of the album.
Conclusion: let’s rock
Will AIP will bring Palantir to “take the whole market”?
CEO Alex Karp is confident:
Attending AIPCon made me an even more confident supporter and proud shareholder after 4 years of maniacal study of the company.
I could appreciate Palantir’s deep technical capability, what is possible to do with AI, and what makes it possible: an extremely talented group of amazing people led by a rockstar CEO.
Let’s rock.
Yours,
Arny
If this article helped you better understand Palantir please hit the ❤️ button and share it!
Reach me:
Disclaimer: The views and opinions expressed above are current as of the date of this document and are subject to change without notice. Materials referenced above will be provided for educational purposes only. None of the above will include investment advice, a recommendation or an offer to sell, or a solicitation of an offer to buy, any securities or investment products.
wow i actually had fun reading this. lots of good info! good work arny!
Fantastic article, and so glad you got to go!