Palantir Technologies’ Post

As someone deeply invested in leveraging technology to drive efficiency and innovation, I found this perspective from Palantir CTO Shyam Sankar incredibly insightful. He reminds us that the real value of AI goes beyond just boosting productivity—it’s about rethinking AI as a type of labor to unlock its full economic potential. This aligns so much with the challenges and opportunities I see in building systems that empower organizations to solve complex problems faster. Highly recommend this for anyone navigating the intersection of AI and enterprise automation!

Exactly the way forward for leveraging agentic concept strategies. The real value lies in using AI not just as a productivity boost, but as an autonomous problem-solving tool for those ‘Day 2’ challenges.

The application of Gen AI as a co-pilot vs automation. Whilst co-pilots are now relatively common, what forward-thinking and innovative enterprises should be looking to do, is fully automating workflows using LLMs. Using an LLM to search through large, complex, disparate data sets is certainly a force multiplier compared to a human users ability to do the same, but it still comes with the bottle-neck of applying that information and the users decision in a workflow or business process. If you can automate the process end to end with appropriate governance, you can empower the user to oversee the workflows and focus only on the high-impact areas that matter to the business. The widespread adoption of this is undoubtedly the future. There is sometimes some cynicism or hesitancy to this discussion, as if it is something to be feared. In my own mind I compare it to adoption of computers and the internet, to think about how different business structure, governance, job roles would have been prior to, and during that transition. It's only a matter of time, and early adopters will be rewarded not only with a business advantage, but the corporate knowledge of applying these tools effectively.

Can't agree more. Edge cases make or break AI systems - just like in self-driving cars, as Shyam explains below. In near future when one person will need to manage 1000s of AI agents, how will you catch these critical edge cases? Continuous improvements will be another nightmare. This is exactly what we're solving at Pype. To learn more visit pypeai.com or DM me.

🎯 "The value is in automation, not just conversation" The future of AI isn't about chatting with robots. It's about transforming work. Key insights from industry leaders: AI co-pilots might make you 50% more productive But FULL AUTOMATION can transform your entire operation The real value? AI as a type of labor Here's what we've learned: The "happy path" is easy Success lies in handling edge cases Continuous feedback is crucial The question isn't whether AI can chat. It's how effectively it can WORK. At AItomization, we're not just building chatbots. We're creating AI workers that handle real business processes 24/7. Ready to move beyond chat and into true automation? #AI #Automation #DigitalTransformation #Future #Innovation #BusinessGrowth

Good talk: AI is a division of labor; we are promoting human managers of AI agents to get automated work done. Drivers of cars have been conducting millions of edge cases for the last 20 years.

Making AI work for us requires a shift in perspective. In this video, Shyam Sankar, CTO of Palantir Technologies, explains why viewing AI as “a new form of labor” is crucial. He discusses the limitations of current AI tools and emphasizes the importance of re-evaluating workflows and human roles - akin to the onset of computers 30 years ago. As a trusted Palantir partner, we understand a key aspect of this is empowering subject matter experts to effectively manage and guide AI agents. This requires clear definitions of human roles, optimized workflows that leverage both human and AI strengths, and tools that facilitate seamless human-AI interaction. Watch the video to hear more of Shyam's perspective on this topic.

How must the workplace adapt to AI in in the next 5 years? AI's economic value will come from not making human's 2x more productive, it'll come humans focusing on evolving AI systems to handle increasingly complex scenarios -- pushing the boundaries of what can be automated and ultimately allowing AI can operate independently on a wide range of tasks, with humans focusing on strategic oversight ("Enterprise Autonomy") I believe this is a big reason why Copilot has been a disaster. To break down this down further... - Why is automation limited by edge cases? Automation works well for standard predictable tasks. However, real-world scenarios often involve unexpected situations or "edge cases" that don't fit the typical pattern. In the context of LLMS, while they can handle a wide range of tasks, it's the unusual or complex situations that require human expertise. What would a tool chain look like? An effective "tool chain" would focus on rapidly identifying & addressing "Day 2 problems"- the complex, unforeseen issues that arise with implementation - Advanced LLM integration for handling routine tasks - Systems for quickly identifying when LLMs encounter edge cases - Rapid learning and adaptation mechanisms to incorporate new knowledge - Interfaces for human experts to easily provide guidance and corrections - Analytics to measure the economic value generated by AI automation - Feedback loops to continuously improve AI performance and autonomy What does it mean to be managers of AI agents? - Defining tasks and objectives for AI systems - Identifying areas to strategically deploy AI automation to maximize economic value - Continuously assessing and redefining the boundary between AI and human tasks between new data and edge cases - Making strategic decisions about when to use AI vs. human labor Shyam Sankar super insightful, thank you. #palantir #AI #Automation #AIagents #Economicvalue #Enterpriseautonomy

Sembix is aligned with the AI value proposition outlined by Palantir’s CTO at DevCon. Get to Day 2 problems faster! https://lnkd.in/gViWQ_MY

This is exactly where things are headed with AI. It’s not the chat. It’s the chat into agents into workflows into autonomous processes with subject matter experts and output-focused development. There are big changes on the way. For example, CRMs are at huge risk to major disruption.