Pradeep Goel

We've solved the first mile of AI, getting a model to respond. The focus now is on the middle mile, building agents that can execute tasks. The next problem is the last mile, how an AI delivers its work back to a human in a way that is intuitive, trustworthy, and actionable. Solving this problem will usher the next wave of productivity.

Compute from cloud providers has been a commodity, priced per gigabyte and per hour, AI is changing that. The most valuable compute is specialized. Compute that is optimized for a specific model, located in a specific geography for data residency, and coupled with specific, proprietary data is no longer a commodity. It's a unique, strategic asset. This is the core of the @Pai3Ai thesis, the future of compute is owning specialized, sovereign nodes that are uniquely configured for your specific AI workload. #AIInfrastructure #Pai3Ai #Compute

The pursuit of AGI is currently framed as a centralized, cloud-based endeavor, requiring unimaginable amounts of centralized compute. This is a failure of imagination. The most resilient and scalable path to AGI is a federated one, where intelligence emerges from the collaboration of millions of specialized, on-device models. Your phone, your laptop, and your car's computer will all host nodes of a larger, distributed intelligence. This is not only more private and secure, but it's also the only architecture that can scale globally without hitting the physical limits of data centers and energy consumption. The brain is a network of neurons. AGI will be the same. #AGI #Decentralization #EdgeAI #FutureOfTech

@fchollet Current systems recombine what exists. Discovery needs models that can form and test hypotheses, they're not independently advancing science

Current AI is a librarian of existing knowledge. Science requires an explorer of the unknown. You don't win a Nobel Prize by staying in the library.

@APompliano Clear observations

Here are 13 things learned after making a big push to integrate AI into our companies: 1. We haven’t replaced a single external SaaS tool with something we built internally. 2. We have refrained from hiring numerous entry level jobs because AI can do the work faster/better/cheaper. 3. The automation provided by AI highlights how much time every person was wasting on tedious tasks daily. 4. Each company is capturing more revenue and each employee is becoming more productive. 5. There is still a bit of apprehension in giving agents full control of machines or systems. 6. There has been no obvious trend in age, gender, or role for those who adopt AI the fastest. More of a mindset than anything. 7. Many non-technical people have started to create software tools or products, which has changed the speed of execution across the companies. 8. One downside is the AI slop across written documents/memos. If humans don’t review the content, it is painful to read and I worry critical thinking gets lost. 9. The implementations of AI are incredible once you get them done, but it is much more difficult to build/implement than most people want to communicate online. Persistence needed! 10. We have walked away from numerous potential small acquisitions because we realized we could build the product ourselves for a fraction of the cost. 11. Our best engineers are invincible now. They produce high quality products at warp speed. Forget 10x engineers, they are 1,000x engineers now. 12. The adoption of AI starts at the top. If the company leader is not constantly asking “how do we automate this?,” it is harder to drive internal change. 13. I am personally working harder than I have in a long time and having more fun than ever. It feels like a moment in time that has to be seized. Overall, I believe AI is underestimated, not overestimated. The worries about SaaS software are probably overblown. The labor market impact is very real and only accelerating. Businesses are fundamentally changing. Start paying attention!

@DrBenTapper1 Compute has a physical footprint people ignore. AI scales digitally, but its constraints are physical, water, energy, land. At some point, deployment decisions will come down to resource allocation

AI data centers can use up to five million gallons of water each day. Am I the only one concerned about how this could pose a serious threat to our farmers and our food supply? What happens when these centers seriously strain our aquifers and dry up irrigation systems? Is it really worth it?

@Cointelegraph We keep framing this as jobs lost vs jobs created, when the issue is timing. New industries take years to absorb the shock of displacement

🚨 INSIGHT: 300M jobs at risk from AI, but infrastructure growth could drive new employment, per Goldman Sachs.

@CryptoMikli He’s right about the structure of the work. But what's more realistic is fewer juniors, not fewer lawyers

Andrew Yang explains why lawyers will be replaced by AI “The first thing that jumped into my mind when you said that was lawyer. Law school applications, last I checked, went up 21% last year, and I would suggest that was a flight to safety, and that stuff’s not safe at all. Lawyering is highly structured. It’s very process oriented. It’s kind of the ideal environment for AI” “I have friends who are partners in law firms who say, ‘Look, I’m giving AI work that would have taken a second or third year associate a week to complete, and it gives it back to me in 20 minutes. So why on earth would I hire a small army of these associates?’”

@PeterDiamandis Abundance shifts competition from access to resources, control of production systems

"Scarcity thinking: 'There are 8 billion people competing for limited resources.' Abundance thinking: 'There are 8 billion minds that could solve the resource problem.' AI + biotech + energy abundance means the competition isn't for the pie. It's to grow the pie 1,000X."

The dominant paradigm for human-computer interaction is still the master-slave relationship of the command line. We command, the computer obeys. AI is enabling a symbiotic operating system, where the relationship is collaborative. In this model, the AI anticipates your needs, suggests actions, and works in parallel with you. It’s not just an assistant, but a cognitive co-processor. This requires a rethinking of interface design away from static menus and towards dynamic, conversational, and context-aware environments. The OS of the future will be one you collaborate with.

The entire conversation around Agentic AI is focused on the agent's brain, it's reasoning, planning, and capabilities. The most important question for any agent is where it lives. An agent is a persistent entity with a memory, an identity, and a set of permissions. Storing that persistent state on a multi-tenant server is a security and privacy nightmare waiting to happen. This is the problem @Pai3Ai was built to solve. A Power Node provides an agent with a sovereign home. It’s a secure, private environment where the agent's memory, its decision-making logic, and its API keys are encrypted and under the user's exclusive control. The agent can venture out onto the public internet to perform tasks, but it always returns to a safe harbor. This architectural choice is more important than the size of the agent's language model. A brilliant brain in an insecure home is a liability. A capable brain in a sovereign home is the most powerful and trustworthy colleague you will ever have. We are building the homes for the next generation of AI. #AgenticAI #AISecurity #Pai3Ai #Privacy

Every autonomous AI agent needs a constitution, a core set of principles encoded into its architecture that governs its behavior. This constitution must define its prime directive, its ethical boundaries, and its transparency obligations. This is a foundational design choice. The quality of an AI's constitution will ultimately determine its trustworthiness and its utility.

AI competition is intensifying globally, with @Xiaomi committing at least $8.7 billion over the next three years to advance its AI capabilities. The investment follows the release of its MiMo-V2-Pro model, which has already processed over 1.5 trillion tokens, signaling strong early adoption among developers. It’s interesting where that investment is going. Chatbot markets are becoming saturated and pricing compressing, companies are moving toward agent-based systems. Models designed to execute tasks with speed and accuracy. These systems consume significantly more compute and tokens, turning usage intensity into a new revenue driver. This is exactly where things are heading with @Pai3Ai. The value is not only in smarter models, but in systems that can act, coordinate, and deliver outcomes. The focus is shifting to execution, reliability, and control as agents become the dominant interface, where AI moves from conversation to real-world impact. #AgenticAI #AIInfrastructure #AICompetition #PAI3

The enterprise IT stack has been built on a foundation of centralized convenience. We traded control for scalability, and data sovereignty for ease of use. This was a reasonable compromise for the applications of the past, but it is a catastrophic failure for the AI of the present. You cannot build a critical infrastructure on rented land. @Pai3Ai is not another AI model or a minor improvement to the cloud. It is the foundational layer of the Sovereignty Stack. Built on a single, non-negotiable principle: your data and your AI models must reside in an environment you control. A @Pai3Ai Power Node is the physical manifestation of this principle. It is the secure, sovereign territory where your most sensitive data can meet the most powerful AI without ever crossing a public boundary. The future of enterprise AI won't be built on public clouds. It will be built on a global fabric of sovereign nodes. #Pai3Ai #DataSovereignty #EnterpriseAI #Infrastructure

We are obsessed with teaching AI to learn, but we are ignoring the more critical problem: teaching it to forget. In a world where AI agents hold persistent memories of our interactions, our data, and our mistakes, the ability to selectively, verifiably, and permanently erase that memory becomes a fundamental human right and a profound technical challenge. Current AI systems are designed for perfect recall. This is not a feature, for human interaction. A trusted colleague doesn't remember every clumsy sentence you've ever said. We need to build a forgetfulness protocol for AI, a way for agents to not just delete data, but to retroactively unwind its influence from their model weights. The systems that can master the art of forgetting will be the ones we can truly trust. #AI #Privacy #DataSovereignty #Ethics

@JonhernandezIA Interesting take

📁 Palmer Luckey, Oculus founder and Anduril CEO, says the AI advantage is not just about technology. It is about speed of deployment. China is integrating AI across its system much faster. Not because it has better models. But because it moves faster.

@rohanpaul_ai AI makes good paths more obvious. But knowing which problems are worth solving is still unevenly distributed

Chamath on how AI agents are making the "10x engineer" distinction disappear because the most efficient "code paths" are now obvious to everyone. Just as AI solved chess and removed the mystery of the best move, AI is doing the same for coding, making the process reductive and removing technical differentiation. "I'm going to say something controversial: I don't think developers anymore have good judgment. Developers get to the answer, or they don't get to the answer, and that's what agents have done. The 10x engineer used to have better judgment than the 1x engineer, but by making everybody a 10x engineer, you're taking judgment away. You're taking code paths that are now obvious and making them available to everybody. It's effectively like what happened in chess: an AI created a solver so everybody understood the most efficient path in every single spot to do the most EV-positive (expected value positive) thing. Coding is very similar in that way; you can reduce it and view it very reductively, so there is no differentiation in code." --- From @theallinpod YT channel (link in comment)

@tomfgoodwin The issue is how quickly narratives form around AI. We skip understanding and jump straight to conclusions, either hype or dismissal

I think AI is an enormously transformative technology. I think it's profound, powerful, magical But the moment people start to talk about "agents will talk to other agents" "brands will have their own agents yours will talk to" All I can see is a tranche of people who don't think. The tech starts to jump the shark All you can see is people who parrot "smart sounding things" There seems to be so little critical thinking in the world. It's either Robots will do everything soon or robots don't work. It's "it's the worst it will ever be" or "this will all get banned and token costs will skyrocket" Can nobody actually ponder on how technology really spreads, or human nature, or incentives, or pretty much any element of history Stop with the daft soundbites

@DeItaone Task level automation is not job level replacement. But it does compress how much human input is needed per role

AI COULD AUTOMATE 93% OF U.S. JOB TASKS A new study finds AI could handle parts of 93% of U.S. jobs, potentially shifting $4.5 trillion in labor costs. Researchers analyzed 18,000+ tasks across 1,000 jobs, with software development, finance, management, legal, and office roles most affected. Cognizant CTO Babak Hodjat notes adoption is uneven, but rapid AI breakthroughs—like multimodal and agentic AI—are accelerating automation. Some physical and care jobs, like construction and healthcare support, see smaller but growing AI impact. Impact doesn’t mean job loss: AI augments human work, improving efficiency and output. Globally, AI could influence $15 trillion in labor value, with its capabilities expanding fast.