AIPCon: practical AI-enabled workflows

Placeholder Image

When I started at Palantir, Foundry 0.1 was over a year away and LLMs like GPT4 or Llama2 weren’t on anyone’s radar. When I left Palantir, Foundry-the-data-platform was real, Foundry-the-operations-suite was being created, and LLMs were the domain of niche R&D labs.

Last week, when I attended AIPCon as a Palantir partner, I saw just how mature Foundry has become and I saw that AIP, Palantir’s LLM-centric AI product announced just earlier this year, is already real.

What do I mean by real? I mean that (a) it’s being used not just in proof-of-concept workflows, but actual production workflows that customers are paying for and deriving value from, and (b) it’s something that customers can explain to each other and discuss cogently without Palantir engineers in the room.

In addition to the live-streamed demos on stage, there were booths where customers were demoing and fielding questions about how their organizations were using Foundry & AIP. As I walked from demo to demo I was struck by how many people were nodding in appreciation of Foundry & AIP’s applicability to their own organizations' challenges.

Those challenges, in my eyes, share a kinship with the challenges organizations have always dealt with. How do we ensure we have the right equipment and staff on hand to care for patients? How do we reduce the time for root-causing equipment service orders and ensure that we tackle the most cost-effective preventative maintenance tasks while that equipment is in the shop? How do we intelligently adjust prices and enable procurement teams with data and tools to negotiate with vendors?

But while the challenges may be familiar, the solutions made possible by AIP are new. And not only new, but exceedingly practical. There was very little flashy AI magic at AIPCon. Rather, there was a lot of clearly valuable workflows enabled and enhanced by what we have collectively begun referring to as AI.

If you’d like to get a taste of AIPCon, here are some highlights from the livestream with timestamps:

  • Shyam’s talk is good if you’re looking for an opinionated primer on LLMs in the enterprise (20min starting at 48:42)
  • Ted’s talk is good if you’re looking for a survey of how different organizations are using AIP (25min starting at 1:08:37)

All of the main-stage customer demos are great

  • Brian Fifarek from Eaton on supply chain (1:53:06)
  • Shailesh Yearram from CPKC on rail network management (2:08:54)
  • Jennifer Duthie from Cintra on smart highways (2:21:31)
  • Michael J. Schlosser from HCA Healthcare on hospital operations (2:31:27)