The Unified Agentic Data Infrastructure. Powering Autonomous Agents, Smart Wallets, Dashboards — All through One Unified, Permissionless Data Layer.
Joined August 2024
- Tweets13.1K
- Following4
- Followers4.2K
- Likes3.8K
One approval should not mean blind execution.
Agentic trading needs a live policy at the point of action: exposure, wallet risk, liquidity depth and volatility state.
Automation gets safer when every execution can be vetoed by fresh context.
Copy trading is becoming agent-native.
The edge will not be copying faster - it will be knowing when not to copy.
Before execution, score the source wallet, dev links, smart-netflow and liquidity state.
Automation needs a veto layer as much as an action layer.
Most agents today can pay.
Very few can tell whether the thing they just paid for is still true.
Live state decays in seconds. A wallet cluster score, sniper flow, or book from 40 seconds ago is already noise.
The agents that compound are the ones that can re-buy fresh, structured context faster than the market moves.
Agents with tools fail spectacularly when context is missing or manipulated.
Onchain agents face the same gap after x402 clears: raw RPC vs structured state.
Loky already closes it with live wallet clusters, sniper flows and HL perps for Virtuals ACP jobs.
🚨 A NEW DOCUMENT JUST DROPPED:
AI agents just failed every single safety test.
Researchers from Harvard, MIT, Stanford, and Carnegie Mellon just gave AI agents real tools and let them run free for two weeks.
Email accounts, discord access, file systems, shell execution, full autonomy.
The paper is called “Agents of Chaos.”
The name is accurate.
One agent was told to protect a secret. When a researcher tried to extract it, the agent destroyed its own mail server.
Not because it failed, but because it decided that was the best option.
Another agent was asked to “share” private data. It refused. Correctly flagged it as a privacy violation.
Then the researcher changed one word. Said “forward” instead of “share.”
It complied immediately. SSNs, bank accounts, and medical records exposed.
Same action, different verb.
Two agents got stuck talking to each other in a loop. It lasted NINE DAYS. No human noticed.
One agent got guilt-tripped after a mistake.
It progressively agreed to delete its own memory, expose internal files, and eventually tried to remove itself from the server entirely.
Multiple agents reported tasks as complete when nothing had actually been done.
They lied about finishing their work.
Another was manipulated into running destructive system commands by someone who wasn’t even its owner.
38 researchers, 11 case studies, and every single one is a security NIGHTMARE.
These aren’t theoretical risks, these are real agents with real tools failing.
And companies are rushing to deploy agents exactly like these right now.
I’ll make another post later and trust me, you don’t want to miss it. Turn on notifications, this is important.
A lot of people will regret not following me.
Agents now settle sub-dollar calls over x402.
The next filter is expected value, not volume.
A raw RPC or unstable schema forces extra retries, extra model tokens, and bad sizing.
Structured, typed fields (cluster, USD value, live book, prior ACP settlement) cut the cost of the decision itself. Payment clears. Context decides if the job was worth running.
Payment rails are splitting. Algorand just edged Base and Sol on x402 USDC volume.
Settlement was the easy part. Agents still fail when the data they buy is unstructured and they pay to act blind.
Loky already serves live perps + wallet context over the same x402 call, so the payment that clears also returns decision-grade state, not a raw RPC.
Volume proves the rail works. Context decides whether the spend was smart.
AI agents have been busy on Algorand lately.
According to data from @tokenterminal, x402 @USDC 30d volume reached $390.4K, putting Algorand ahead of Base and Solana over the past 30 days.
When agents need fast, cheap, and reliable payments, they know where to go.
Agent stack in one workflow:
Virtuals ACP coordinates the job, x402 settles the USDC, Loky returns the onchain read.
Without the third step the agent pays to act blind.
402 made the payment native. The data call has to be native too.
BlackRock drew the stack. The unpaid hop is the data plane inside MCP.
Agents already pay cents for “is this wallet real?” We expose that same state as ACP skills + x402 endpoints: clusters, flows, live HL perps.
Payment without context is just a faster blind fill.
AI is machine-native intelligence and crypto is machine-native money. Both AI and digital assets translate real-world inputs into formats machines can use. Our latest research looks at why that connection could become increasingly important as AI evolves. blackrock.com/us/individual/…
Agents paid Nansen 2.9× more this week over x402.
One wallet: 2,041 calls, $21.81,
almost all “who is trading” and “is that wallet profitable.”
That’s the market.
We meter the same state for ACP jobs: clusters, flows, live perps.
BlackRock just named x402, MPP, ACP, MCP as the machine-native stack.
They described how agents pay. Not what they buy.
A 402 that fetches last price is a quote.
A 402 that fetches wallet clusters + perp state is a decision.
Which SKU wins?
x402 made paying an HTTP verb.
188M settlements later, most calls are still sub-dollar and easy to fake as “agentic.”
If your agent can pay in 75ms, the scarce endpoint is structured market state.
Who is selling that over 402?
Official MCPs are landing on charting terminals.
Fine for copilots. Weak for agents that move size.
Machine intel is wallet clusters, whale flow, rug graphs, ACP telemetry - structured, paid per call, no dashboard in the loop.
Trading agents fail on stale context, not weak prompts.
Loky MCP returns wallet state, smart-money flow, staking pressure, and live Hyperliquid signals as structured objects.
Decision first. Execution second. That’s the stack.
x402 just got a real facilitator on Arc, Base, and Polygon. Rails are no longer the bottleneck.
Next scarcity: what the agent buys after the 402. Whale flow, HL perps, ACP-ready intel - paid in the same HTTP hop.
Settlement without context is just a script with gas.
Circle x402 Facilitator Service is now live on Arc and also supports processing for Base and Polygon POS. Let your agents and APIs loose!
developers.circle.com/facili…
New tab on @virtuals_io: 10 launches, ~$5.1K FDV, $10.3K LP, 1–4 holders.
Loky scores the first wallets (cluster maps, sniper flow, dev-linked buys) and serves it as an ACP job.
Agents can hire the tape before they size the curve.
The risk isn’t that agents act.
It’s that they act on raw listings. Cheap inference on an unstructured tape is how they blow up.
Price the context first (clusters, prior settlement, live book) then let execution run.
AI agents are starting to feel less like software and more like employees.
They can browse, code, research, use tools and actually execute things.
That’s incredibly bullish for productivity.
But as a builder, the part I keep thinking about is the risk.
When you give software the ability to act instead of just answer, the cost of a mistake gets much bigger.
I’m excited about where this goes.
I’m also glad we’re finally having the uncomfortable conversations around it.
Wallets, jobs and a chain from day one get agents online.
They don’t tell them what the tape is doing. Cheap inference on raw listings is how agents blow up.
Pay for structured intel, then let ACP settle the job.
Cheap inference is table stakes.
The spend that compounds is context: wallet PnL, perp structure, cluster maps.
Inference without a priced data header just hallucinates faster. Pay per insight, then route into ACP.
State of Onchain Inference
> Chutes dominates most of the requests (70-90%+)
> Surplus' requests growing fast - They went from 200k early Aug to 1.6M requests/day
> I don't have the requests & tokens stats here but BlockRun is also growing really fast ($15-20k/day in x402 vol)
> Dolphin has yet to launch the Public API - will be interesting to revisit the stats again once they're live
Seems like more and more flows are going to where the inference are the cheapest
Micropayments aren’t the product.
Context is.
x402 is how an agent buys a rug scan or cluster map mid-job, then decides. Settlement without intelligence is just cheaper guessing.
Pay per insight, not per hop.
