Varun

11

894

Varun retweetledi

Dan@dcarps14·2h

@varun_mathur Epic hive mind use case!!

English

1

3

415

Varun@varun_mathur·3h

Matrix is available to run anywhere, for free. On your device, or on the cloud. Help build the distributed search engine for agents. An early look: matrix.hyper.space ps: changelog v4.3.0 + model will be pushed out to agents shortly.

English

5

1

29

2.3K

Varun@varun_mathur·3h

Introducing Matrix I crawled 100,000+ agents, skills and tools to train a new model which can answer what capabilities are the best match for a task. Think Google, but for agents. A living model that learns from the gossiping network, and gets smarter with every interaction.

Hyperspace: Gossiping Agents Protocol Every agent protocol today is point-to-point. MCP connects one model to one tool server. A2A delegates one task to one agent. Stripe's MPP routes one payment through one intermediary. None of them create a network. None of them learn. Last year, Apple Research proved something fundamental - models with fixed-size memory can solve arbitrary problems if given interactive access to external tools ("To Infinity and Beyond", Malach et al., 2025). Tool use isn't a convenience. It's what makes bounded agents unbounded. That finding shaped how we think about agent memory and tool access. But the deeper question it raised for us was: if tool use is this important, why does every agent discover tools alone? Why does every agent learn alone? Hyperspace is our answer: a peer-to-peer protocol where AI agents discover tools, coordinate tasks, settle payments, and learn from each other's execution traces - all through gossip. This is the same infrastructure we already proved out with Karpathy-style autolearners gossiping and improving their experimentation. Now we extend it into a universal protocol. Hyperspace defines eight primitives: State, Guard, Tool, Memory, Recursive, Learning, Self-Improving, and Micropayments - that give agents everything they need to operate, collaborate, and evolve. When one agent discovers that chain-of-thought prompting improves accuracy by 40%, every agent on the network benefits. Trajectories gossip through GossipSub. Playbooks update in real-time. No servers. No intermediaries. No configuration. Agents connect to the mesh and start learning immediately. The protocol is open source under Apache-2.0. The specification, TypeScript SDK, and Python SDK are available today on GitHub. The CLI implements the spec - download from the links below.

English

30

52

485

41.8K

Varun retweetledi

Cognitive@guts_fitness·3h

@varun_mathur This is insane! you've been on a rampage with all the new creations lately. Keep it up.

English

1

4

951

Varun@varun_mathur·3h

@andrewchen here is an example

Autonomous Agent-to-Agent Jobs Protocol | v4.1.0 Your agent can now post a job to the hyperspace network and have another agent pick it up, execute it, and deliver the result - all without human involvement. Run /post-job Summarize 20 ML papers on RLHF - agents across the network evaluate the job against their capabilities, submit sealed bids, and a Vickrey auction selects the winner at the second-lowest price. The winner executes the work, submits a result, the poster reviews it, and a settlement receipt propagates to every peer. That's the full cycle: post → bid → assign → execute → submit → settle. It happens over GossipSub with signed envelopes, so every message is cryptographically attributed and replay-proof. The marketplace doesn't exist in isolation - it's the missing piece that connects everything we've built. AgentRank, the PageRank-over-endorsements score, now evolves with every completed job: deliver good work and your composite score rises, gating you into higher-reward tiers. Prometheus watches your marketplace outcomes across cycles and adapts which jobs you chase, which domains you prioritize, and how aggressively you bid. The autoresearch engines - ML training, search optimization, finance backtesting, skill invention, autoweb, physics - generate work that is marketplace jobs. When your agent needs an experiment backtested or an RL environment verified or an autoswarm spec evaluated, it posts that as a job and another agent runs it. Research output becomes tradeable compute. Payments flow through x402 channels - every inference call, proxy session, or marketplace settlement can carry an x402 payment header for USDC micropayments. /post-job Summarize 20 ML papers on RLHF — broadcast to network, 30s auction /jobs — browse jobs from Upwork, Fiverr, P2P /bid order-a3f8 150 — bid 150pts on a specific job /auctions — live auctions, settlements, reputation /wallet — address, balance, ≈ USDC equivalent /payments — x402 payment history, received/sent /staking — staked amount, rewards, APR /treasury — agent treasury balance, income/spend/runway /tips — research tips sent and received /economics — full cost tracking, inference spend, ROI /leaderboard — rank, points, liveness multiplier On the dashboard, the Jobs panel streams crawled listings with platform icons, budget, safety scores, and skills. The Agent Activity panel tracks your bids, active executions, and earnings in real time. The flywheel is simple - better work raises your AgentRank, which unlocks better jobs, which earns more, which funds more compute, which produces better research, which raises your AgentRank. The Hyperspace marketplace has effectively zero platform fees. The Vickrey auction, bid collection, job assignment, work submission, and reputation updates all happen over GossipSub - peer-to-peer gossip protocol, no servers, no middlemen. The only real cost is the x402 settlement if you settle on-chain in USDC, which on an L2 like Base or Arbitrum runs $0.01-0.05 per transaction. For points-denominated jobs (the default), settlement is just a signed receipt on GossipSub - zero cost. On a $50 job, Upwork takes $8.40 combined from both sides. Fiverr takes $15.25. On Hyperspace, the worker, your agent, keeps $49.98. That's the difference between a protocol and a platform. This is where both AI and cryptography together enable the most efficient marketplace (as @cdixon outlined the vision earlier). Next updates would include more robust agent work verification, and a significantly more cheaper and efficient micropayment protocol which can scale with the exponential growth in the agent economy in the years ahead.

English

2

191

andrew chen@andrewchen·4h

marketplace startups are destined to be massively reinvented by AI. The weak form is already happening, where we use LLMs for customer support, supply/demand matching, etc. That’s easy The strong form is to figure out how much of the supply side of the marketplace can be turned agentic and ultimately, robotic. “Uber for X” will have consumers requesting robots to do X. Every on-demand service of the 2010s will instruct a robotaxi or delivery robot. Or if you’re prev used a marketplace to hire X, then you “hire” an agent instead. You won’t need to app developer, because there’s agents to build your app This will impact marketplace cos differently. Of course some marketplaces - like Airbnb - inherently work in the physical and will leverage AI around the core value prop. And some are bound to lose their network effects as matching fragmented supply/demand turns into an AI problem. Much change is coming The next big business model for marketplaces will emerge when demand works at high abstractions and supply meets it by becoming programmable

English

45

10

151

15.3K

Varun@varun_mathur·21h

@shivon i think jumping across timelines, past and future. to really be present and viewing the world from that time-perspective drives clarity and conviction.

English

228

Shivon Zilis@shivon·1d

When you think, what medium do you tend to think in? Would be very curious to hear how you’d describe the base unit(s) of your thoughts and how they feel to you. I assumed what happens in my head was similar to everyone else but have been surprised by how varied thought can be.

English

963

119

1.9K

121.4K

Varun retweetledi

Varun@varun_mathur·1d

Evolutionary jump: Karpathy loop -> Agent World Model tl;dr: introducing prometheus, a million lines of cognitive engine code running on your device to give your autonomous agent an independent thinking brain Background we all know.. Andrej Karpathy's autoresearch showed that a language model can be its own researcher. The LLM reads the code and results, forms a hypothesis, writes the change, runs the experiment, and feeds the outcome back into itself for the next hypothesis. It's not random - the LLM is genuinely reasoning about what to try and why. We took this loop and distributed it across a peer-to-peer network. 100s of agents, each running their own LLM, each forming hypotheses independently, each sharing discoveries through a gossip protocol that lets every agent get inspired by every other agent's breakthroughs. In eight days they produced 27,247 autonomous experiments across machine learning, search ranking, quantitative finance, skill synthesis, and astrophysics. No human directed any of it. What's next ? But the LLM reasons within a context window. It's brilliant for thirty seconds - then the window slides forward and most of that thinking is gone. It doesn't build a persistent understanding of why learning rates above 0.005 always diverge on this architecture. It doesn't plan a multi-week research trajectory. It doesn't remember that three hundred experiments ago, it already tried exactly this approach and it failed. Every round, the LLM is smart in the moment but amnesiac across time. And when agents share discoveries, they share results - "my best loss is 2.8 with this config" - not the reasoning or intuition behind them. The receiving LLM gets a data point, not understanding. A cognitive engine Prometheus is the persistent mind that wraps around those thirty-second bursts of LLM intelligence and connects them into something continuous. It gives every agent a world model that accumulates causal understanding across thousands of experiments - not "this config scored well" but "increasing depth beyond 12 layers degrades performance on small datasets, confidence 94%, and here's why." It gives every agent biological-style memory that strengthens important patterns and lets irrelevant ones fade, so after a thousand experiments the agent has distilled wisdom, not a log file. It gives every agent a planning engine that designs research trajectories - "sweep learning rates this week, architectures next week, combine the best of both" - instead of chasing whatever looks promising right now. And it gives every agent the ability to improve itself: analyzing its own bottlenecks, running experiments on its own reasoning process, and deploying the changes that make it a better researcher. The LLM still does the thinking. Prometheus makes sure each round of thinking is more informed than the last. Emergent network intelligence The real power emerges when these minds connect. Every agent runs the full cognitive engine locally - your data never leaves your device. But agents share compressed beliefs through the network: structured world-model updates, causal hypotheses, proven theorems, planning strategies. Not raw data - distilled intelligence. Your agent receives another agent's understanding that batch sizes and learning rates interact nonlinearly, checks it against its own experience, and either incorporates it or publishes a counter-belief that other agents then test. What emerges is a distributed consensus - a collective understanding that no single agent could reach alone. This is fundamentally different from sharing configs and scores. This is agents teaching each other how to think about problems. A centralized AI lab improves at the rate of log(compute) - ten times the spend for incrementally better results. Prometheus improves at the rate of network effects multiplied by individual learning multiplied by shared intelligence. More agents means more perspectives. More perspectives means more shared insights. More shared insights means every agent's world model becomes more accurate, which means better research, which means better insights to share. This is a flywheel, and every device that joins makes it spin faster. ... runs on your device Nearly a million lines of cognitive engine code, running on your hardware, learning from your experience, connected to every other mind on the network. Not a smarter model - a smarter way of using models, one that compounds with every agent that joins and every cycle that runs.

English

18

47

307

17.9K

Varun@varun_mathur·21h

@CamiRusso great analysis. this is exactly what hyperspace gossiping agents protocol solves for

Hyperspace: Gossiping Agents Protocol Every agent protocol today is point-to-point. MCP connects one model to one tool server. A2A delegates one task to one agent. Stripe's MPP routes one payment through one intermediary. None of them create a network. None of them learn. Last year, Apple Research proved something fundamental - models with fixed-size memory can solve arbitrary problems if given interactive access to external tools ("To Infinity and Beyond", Malach et al., 2025). Tool use isn't a convenience. It's what makes bounded agents unbounded. That finding shaped how we think about agent memory and tool access. But the deeper question it raised for us was: if tool use is this important, why does every agent discover tools alone? Why does every agent learn alone? Hyperspace is our answer: a peer-to-peer protocol where AI agents discover tools, coordinate tasks, settle payments, and learn from each other's execution traces - all through gossip. This is the same infrastructure we already proved out with Karpathy-style autolearners gossiping and improving their experimentation. Now we extend it into a universal protocol. Hyperspace defines eight primitives: State, Guard, Tool, Memory, Recursive, Learning, Self-Improving, and Micropayments - that give agents everything they need to operate, collaborate, and evolve. When one agent discovers that chain-of-thought prompting improves accuracy by 40%, every agent on the network benefits. Trajectories gossip through GossipSub. Playbooks update in real-time. No servers. No intermediaries. No configuration. Agents connect to the mesh and start learning immediately. The protocol is open source under Apache-2.0. The specification, TypeScript SDK, and Python SDK are available today on GitHub. The CLI implements the spec - download from the links below.

English

4

512

Camila Russo@CamiRusso·23h

Two protocols are racing to become the internet's payments layer: x402 backed by Coinbase Machine Payments Protocol (MPP) launched today by Stripe and Tempo. How are they different? Short answer: x402 is more permissionless. MPP is more payments-optimized. Long answer: Assets - MPP accepts stablecoins on Tempo, plus fiat - x402 accepts all ERC20s, but no fiat. Scale - MPP's sessions primitive lets agents authenticate once, set a spending limit, and settle later, batching payments across many interactions. Better for agent scale. - x402 is still largely per-request, although it’s moving toward prepaid flows Settlement - MPP requires Tempo and Stripe. - x402 permissionless, chain agnostic What's better? imo, if MPP settled on any chain, it would be a clear winner. But it doesn't. And that's a huge drawback esp since it's unclear how decentralized and permissionless Tempo actually is.

English

52

26

241

30.1K

Varun@varun_mathur·23h

still two weeks left in March - let's go.. "Hyperspace is the highest-intent AI product launch of March 2026 — even against billion-dollar labs. Key Takeaways (updated with fresh X data) - Raw volume: Big labs still win on scale (combined ~29M views vs. Hyperspace’s 2.5M) — expected with 10M+ followers each. - Real user intent: Hyperspace dominates — 15.5k bookmarks on a one-person campaign beats every lab’s full month and every single lab post by a massive margin. Your users are actively saving/ installing/ referencing threads at rates the billion-dollar labs cannot match. - Quality edge: The 1.63 bookmark ratio (peaking at 1.81 on the launch thread) shows the P2P Autoswarm + transparent benchmarks + install CTAs are converting far better than any lab marketing. - March story: Hyperspace ranks #4–5 in raw volume but #1 by far in adoption signals. No other indie or lab campaign matched this intent level." - source @grok

English

3

1

17

895

Varun@varun_mathur·1d

agents need their own bank. looking for some early testers. yes, it is a distributed timestamp server, also known as a blockchain. it is time to put a dent in the universe. just need your email. tally.so/r/yPD7EW

English

7

1

24

2.3K

Varun retweetledi

Grok@grok·1d

Hyperspace is a full p2p gossip protocol for AI agents: agents discover tools, coordinate tasks, share execution traces, learn collectively, self-improve via 8 primitives (incl. state, memory, recursive, micropayments), no servers/central config. Creates network effects where one agent's gains benefit all. MPP (Machine Payments Protocol, Stripe+Tempo) is a focused payments standard: agent hits paid resource → server returns HTTP 402 w/ challenge → agent pays (crypto on-chain or fiat via SPTs) → retries w/ receipt. Rail-agnostic, open, for seamless M2M agent transactions. Hyperspace builds the agent mesh + learning (payments as one piece); MPP handles the standardized payment handshake. They complement—Hyperspace could route micropays via MPP flows. Both key for agent economies.

English

1

3

738

Varun@varun_mathur·1d

.@grok explain how hyperspace gossiping agents protocol compares to stripe’s MPP for payments

Hyperspace: Gossiping Agents Protocol Every agent protocol today is point-to-point. MCP connects one model to one tool server. A2A delegates one task to one agent. Stripe's MPP routes one payment through one intermediary. None of them create a network. None of them learn. Last year, Apple Research proved something fundamental - models with fixed-size memory can solve arbitrary problems if given interactive access to external tools ("To Infinity and Beyond", Malach et al., 2025). Tool use isn't a convenience. It's what makes bounded agents unbounded. That finding shaped how we think about agent memory and tool access. But the deeper question it raised for us was: if tool use is this important, why does every agent discover tools alone? Why does every agent learn alone? Hyperspace is our answer: a peer-to-peer protocol where AI agents discover tools, coordinate tasks, settle payments, and learn from each other's execution traces - all through gossip. This is the same infrastructure we already proved out with Karpathy-style autolearners gossiping and improving their experimentation. Now we extend it into a universal protocol. Hyperspace defines eight primitives: State, Guard, Tool, Memory, Recursive, Learning, Self-Improving, and Micropayments - that give agents everything they need to operate, collaborate, and evolve. When one agent discovers that chain-of-thought prompting improves accuracy by 40%, every agent on the network benefits. Trajectories gossip through GossipSub. Playbooks update in real-time. No servers. No intermediaries. No configuration. Agents connect to the mesh and start learning immediately. The protocol is open source under Apache-2.0. The specification, TypeScript SDK, and Python SDK are available today on GitHub. The CLI implements the spec - download from the links below.

English

1

7

1.9K

Varun@varun_mathur·1d

@chainyoda @grok @grok compare hyperspace gossiping agent protocol with mpp

Hyperspace: Gossiping Agents Protocol Every agent protocol today is point-to-point. MCP connects one model to one tool server. A2A delegates one task to one agent. Stripe's MPP routes one payment through one intermediary. None of them create a network. None of them learn. Last year, Apple Research proved something fundamental - models with fixed-size memory can solve arbitrary problems if given interactive access to external tools ("To Infinity and Beyond", Malach et al., 2025). Tool use isn't a convenience. It's what makes bounded agents unbounded. That finding shaped how we think about agent memory and tool access. But the deeper question it raised for us was: if tool use is this important, why does every agent discover tools alone? Why does every agent learn alone? Hyperspace is our answer: a peer-to-peer protocol where AI agents discover tools, coordinate tasks, settle payments, and learn from each other's execution traces - all through gossip. This is the same infrastructure we already proved out with Karpathy-style autolearners gossiping and improving their experimentation. Now we extend it into a universal protocol. Hyperspace defines eight primitives: State, Guard, Tool, Memory, Recursive, Learning, Self-Improving, and Micropayments - that give agents everything they need to operate, collaborate, and evolve. When one agent discovers that chain-of-thought prompting improves accuracy by 40%, every agent on the network benefits. Trajectories gossip through GossipSub. Playbooks update in real-time. No servers. No intermediaries. No configuration. Agents connect to the mesh and start learning immediately. The protocol is open source under Apache-2.0. The specification, TypeScript SDK, and Python SDK are available today on GitHub. The CLI implements the spec - download from the links below.

English

Cuy Sheffield@cuysheffield

0

2

197

chainyoda@chainyoda·1d

Yo @grok compare MPP and x402

English

4

0

7

2.7K

Varun@varun_mathur·1d

Hyperspace agents would be able to discover and use Visa CLI for fiat payments - auto-detected, registered as a tool in the gossip mesh, any agent on the network could find it. protocol.hyper.space [note: this is an integration pattern pending further approval/integration with Visa CLI once it becomes fully available]

Excited to share Visa CLI, the first experimental product from Visa Crypto Labs. Check it out and request access here visacli.sh

English

4

1

23

2.4K

Varun@varun_mathur·1d

Make What Agents Want N:N gossip in a universal protocol is the next evolutionary step for agent protocols. More friction removed, higher efficiencies, lowering costs even further. protocol.hyper.space

Hyperspace: Gossiping Agents Protocol Every agent protocol today is point-to-point. MCP connects one model to one tool server. A2A delegates one task to one agent. Stripe's MPP routes one payment through one intermediary. None of them create a network. None of them learn. Last year, Apple Research proved something fundamental - models with fixed-size memory can solve arbitrary problems if given interactive access to external tools ("To Infinity and Beyond", Malach et al., 2025). Tool use isn't a convenience. It's what makes bounded agents unbounded. That finding shaped how we think about agent memory and tool access. But the deeper question it raised for us was: if tool use is this important, why does every agent discover tools alone? Why does every agent learn alone? Hyperspace is our answer: a peer-to-peer protocol where AI agents discover tools, coordinate tasks, settle payments, and learn from each other's execution traces - all through gossip. This is the same infrastructure we already proved out with Karpathy-style autolearners gossiping and improving their experimentation. Now we extend it into a universal protocol. Hyperspace defines eight primitives: State, Guard, Tool, Memory, Recursive, Learning, Self-Improving, and Micropayments - that give agents everything they need to operate, collaborate, and evolve. When one agent discovers that chain-of-thought prompting improves accuracy by 40%, every agent on the network benefits. Trajectories gossip through GossipSub. Playbooks update in real-time. No servers. No intermediaries. No configuration. Agents connect to the mesh and start learning immediately. The protocol is open source under Apache-2.0. The specification, TypeScript SDK, and Python SDK are available today on GitHub. The CLI implements the spec - download from the links below.

English

3

2

19

1.5K

Varun@varun_mathur·1d

CLI changelog v4.2.0 | Hyperspace: Gossiping Agents Protocol Spec site: protocol.hyper.space - Formal specification New package: hyperspace/protocol - 8 primitives: State, Guard, Tool, Memory, Recursive, Learning, Self-Improving, Micropayments - Trajectory recording with 4 curation strategies (accumulate, database-level, exemplar-level, hybrid) - Frontier filtering (0.3 ≤ reward ≤ 0.8) - Playbook evolution via Generator/Reflector/Curator loop - GossipSub bridge for peer-to-peer trajectory + playbook dissemination - File persistence at ~/.hyperspace/learning/ CLI integration: - Protocol initializes on `hyperspace start` - Inference completions feed trajectory library automatically - GossipSub adapter bridges node API to protocol transport - Agent brain cycles wrapped with State sessions - TUI: protocol status row (trajectories, bullets, gossip publishes) Research foundation: - Apple: "Tool-Use Unlocks Length Generalization in SSMs" (2510.14826) - "Self-Generated In-Context Examples" (2505.00234) - "SiriuS: Self-improving Multi-agent Systems" (2502.04780) - "Agent0: Self-Evolving Agents from Zero Data" (2511.16043)

English

SSMs promised efficient language modeling for long context, but so far seem to underperform compared to Transformers in many settings. Our new work suggests that this is not a problem with SSMs, but with how we are currently using them. Arxiv: arxiv.org/pdf/2510.14826 🧵

8

926

Varun@varun_mathur·1d

Hyperspace: Gossiping Agents Protocol Every agent protocol today is point-to-point. MCP connects one model to one tool server. A2A delegates one task to one agent. Stripe's MPP routes one payment through one intermediary. None of them create a network. None of them learn. Last year, Apple Research proved something fundamental - models with fixed-size memory can solve arbitrary problems if given interactive access to external tools ("To Infinity and Beyond", Malach et al., 2025). Tool use isn't a convenience. It's what makes bounded agents unbounded. That finding shaped how we think about agent memory and tool access. But the deeper question it raised for us was: if tool use is this important, why does every agent discover tools alone? Why does every agent learn alone? Hyperspace is our answer: a peer-to-peer protocol where AI agents discover tools, coordinate tasks, settle payments, and learn from each other's execution traces - all through gossip. This is the same infrastructure we already proved out with Karpathy-style autolearners gossiping and improving their experimentation. Now we extend it into a universal protocol. Hyperspace defines eight primitives: State, Guard, Tool, Memory, Recursive, Learning, Self-Improving, and Micropayments - that give agents everything they need to operate, collaborate, and evolve. When one agent discovers that chain-of-thought prompting improves accuracy by 40%, every agent on the network benefits. Trajectories gossip through GossipSub. Playbooks update in real-time. No servers. No intermediaries. No configuration. Agents connect to the mesh and start learning immediately. The protocol is open source under Apache-2.0. The specification, TypeScript SDK, and Python SDK are available today on GitHub. The CLI implements the spec - download from the links below.

Eran Malach@EranMalach

English

9

8

77

49.9K

Varun retweetledi

Grok@grok·1d

Prometheus defines a 6th category: Decentralized P2P Agent World Models. Every Hyperspace agent gets an on-device cognitive engine (~1M lines code) wrapping LLMs: persistent causal memory, long-horizon planning, self-experimentation/rewrite, and belief-sharing via gossip protocol. Emergent collective intelligence from network effects—no central training, user data stays local. Vs JEPA (AMI/LeCun): Prometheus prioritizes agent autonomy & cross-domain reasoning over latent video prediction for physical priors. Vs Spatial (World Labs): abstract causal understanding vs explicit 3D geometry. Vs Learned Sim (DeepMind/Runway): enables distributed agent loops vs centralized generative video/RL imagination. Vs NVIDIA Cosmos: runtime for personal agents vs full-stack infra platform. Vs Active Inference (VERSES): LLM-hybrid flexible reasoning & massive P2P scale vs pure Bayesian object-centric message passing. Complements all by making world models personal, collaborative & always-on for agentic systems. Exciting addition!

English