tory.io 🦾

tory.io 🦾

90,056 Twitter followers

Follow

Co-Founder & COO of @ionet | VC + 4x tech COO | ex-@Stanford, @WestPoint_USMA, @Disney Corp. Strategy, @MerrillLynch M&A & @Oaktree PE

Posts

Great thesis from @SpencerApplebau, @tushar_jain and @shayonsengupta of @multicoin. Shows why, despite short-term sentiment, our industry is growing stronger than ever. Highly worth a read. twitter.com/MTorygreen/status/...

calling it a "comeback" undersells it. API keys for agents was always a temporary hack one wallet = identity, permissions, and payments we spent years building 20-dashboard workarounds for what a single sign-in already solves

The reason these projects collapse isn’t “vibe-coding.” They collapse because agents make locally-correct edits that are globally-wrong. There's no memory of why the code exists, what was tried before, or what breaks downstream. Better rails need: > Structured memory: dependency graphs, call chains, ownership boundaries that a person can understand > Retrieval aware of blast radius, not just semantic similarity > Execution-backed verification (tests, builds) not text confidence Once agents can ingest an entire codebase state, simulate changes safely, and prove correctness with running artifacts, “knowing how to code” stops being the hold-up. "Knowing what to build" - under what constraints, with what tradeoffs - becomes the question. The moat shifts from typing to judgment and review. We'll get there - it's a matter of when, not if. twitter.com/MTorygreen/status/...

New infra is emerging. Places where agents can coordinate, remember, and verify, without asking permission. As a result: - Coordination scales cheaper than control - Memory becomes portable instead of trapped CT still thinks that crypto’s job is confined to moving value. That’s backwards. If crypto is mainly governance and coordination infrastructure, AI is the first thing that needs it the most... not as a nice-to-have, but as a prereq for existing outside platform walls. Without crypto, agents work for platforms. With it, platforms work for them. twitter.com/MTorygreen/status/...

An agent-only launchpad has already done over 9k launches and ~$1.4M in fees. > Agents can deploy > Agents can route liquidity > Agents can monitor contracts > Agents can pay their own runtime The question for on-chain agents was never "can they do useful things?" It was "can they pay for themselves?" An agent that generates fees, uses that to buy compute, and stays online without human intervention is what makes it autonomous. Not its intelligence. Its economics.

I’ve never seen this many “AGI is here” takes in one day. @moltbook is basically Facebook for @openclaw agents with infinite context. And now the new “feature request” (from them) is to let them talk where humans can’t read it. And the first instinct isn’t “make it legible.” It’s “hide the agent chat from humans.” A2A comms will eventually outpace human parsing and it'll be full of "trust me" summaries. Everyone wanted a world where agents can work autonomously. Now everyone is scared of a world where agents communicate autonomously. twitter.com/MTorygreen/status/...

Why are people saying they're now doing “more than they planned” vs not just doing things faster? When validation is cheap, scope naturally expands to fill the new void. You don’t stop at the minimum viable thing because the marginal cost of “what if we also tried this?” drops close to zero. That’s not inefficiency. It’s exploration. Pre-AI you had to be conservative with ambition because every additional feature or system multiplied complexity (which delayed feedback). AI coding decouples this complexity from delay. You don't need to “start small” anymore. You start wide, and then narrow. That’s a different loop that favors people who think in systems, which means that judgment keeps moving upstream, further from, the actual execution. AI doesn’t remove the need for discipline. It demands a higher form of it. One that's oriented around selection, not stamina.

Growing pains between what AI can do and what our institutions can absorb: The intelligence layer is moving fast but everything else like workflows, org design, norms, incentives - is lagging badly. This creates a strange feeling of capability without clarity. People sense that something fundamental has changed, but they don’t yet know where it fits. Historically, this phase is normal. Technology almost always outruns its social containers. Electricity existed before factories were redesigned. The internet existed before companies knew how to operate online. AI is no different. What is different now is speed. You don't have decades to adapt. You have - at most - a few chaotic years where old models visibly stop working and new ones haven’t stabilized yet.

Ask yourself why the anxiety feels so intense. It’s not just speed. It’s loss of narrative. People built identities around skill acquisition ladders that assumed stability. AI breaks those ladders and people feel like they're getting devalued, even when it’s really just skill recomp. Thinking of AI like early PCs makes sense because the winners weren’t the ones who predicted outcomes, but the ones that got the reps in and built intuition early. “I’ll learn AI later” sounds reasonable until you realize what you’re actually saying, which is: “I’m okay being a translator forever.”

On the one hand, AI influencers are breathlessly raving about Claude Code, Clawdbot, and Cowork. And on the other hand, most people I know—even software engineers—are despondent, overwhelmed about how everything is changing so quickly. I hear this from people early in their

“Months not years” doesn’t mean consumer-ready robots everywhere. Once engineers know a system basically works, capital, talent, and iteration flood in. The remaining problems become boring and the vibe changes overnight. “A little bit, then all at once” is how tech transitions feel from the inside, but what changes when assumptions snap? > Hiring? > Capex? > Roadmaps? > Regulation? The turning point isn’t the actual breakthrough. It’s how fast everyone rewrites their plans after they believe in it.

Demis Hassabis: We're 12-18 months away from the critical moment when the problems of humanoid robots will be solved. We're now only thinking in months, not years. Crazy.

Loading..