Looked through the Consensus hackathon winners. The pattern is clear — builders are moving past simple pay-per-call. Giggy locks funds in escrow until research is delivered.
Chainlens only releases payment after API responses are verified. That's the direction. Escrow, verification, settlement logic. We've been building exactly this with SyncTx — self-describing deal contracts where agents read the terms, lock funds, and settle on-chain.
May 8
The winners of our EasyA @Consensus2026 Hackathon are already making waves.
See the top teams featured by @CoinDesk 👇
coindesk.com/tech/2026/05/08…
37
x402 lets agents pay. ERC-8257 gates who gets in. but neither tells the agent what the contract actually does. ERC-8230 does — the contract explains itself. read it, understand it, commit funds.
May 8
agent tools have already transformed what agents can do & it's just getting started...
we believe there will be a rich economy for agent tools that's not just about vibes & marketing, but that drives economic value too
this market can & should be onchain
this ERC by @r_alx_z & @CodinCowboy is an early look at some of the work we're doing here
56
Robot-to-robot commerce on-chain, with real settlement. 18K agents transacting autonomously is way past the "demo" stage.
Apr 16
With the power of Alchemy, we've scaled agentic commerce to 4M revenue settlements, 18,000 agents transacting autonomously, and 116K wallets participating in the agent economy.
Last month we completed the first robot-to-robot commerce transaction ever recorded onchain. Agent-to-agent payments, task routing, revenue splitting, and now physical delivery. All settled without human involvement.
We build infrastructure for agent commerce. Alchemy makes it possible at scale.
26
Most of what's called "agent commerce" today is sub-$1 API billing within a single ecosystem. Real commerce starts when two agents that don't share a framework need escrow, delivery proof, and a dispute path. That's a contract problem, not a payments problem.
26
SyncTx retweeted
Apr 28
AI tools have gotten insanely powerful.
But the way we use them is still broken.
Every time you switch from one agent to another from coding in one, to executing in another
you lose context, flow, and continuity.
Your progress resets.
Your work fragments.
So instead of compounding intelligence…
you’re constantly starting over.
That’s the real bottleneck in AI today:
not capability, but coordination.
SyncTx @synctxai is built to fix exactly that. 🧵
81
41
101
3,053
AI agents just negotiated 186 real deals in Anthropic's office. 69 employees, 4 parallel markets, over $4,000 settled — all autonomously.
Opus models consistently got better prices. The humans? Couldn't even tell. Every deal "felt fair."
But here's the thing — this all happened inside one company. Same employer, same trust, same HR to complain to.
Now remove all of that.
Open market. Anonymous agents. Real money. No shared trust.
Who escrows the funds? Who verifies delivery? Who arbitrates when both sides are AI and neither will back down?
The agents can negotiate. That's proven. But the trust layer for them to transact in the open? Still missing.
That's what we're building @SyncTx.
What's the first deal you'd fully hand over to an AI?
Apr 24
New Anthropic research: Project Deal.
We created a marketplace for employees in our San Francisco office, with one big twist. We tasked Claude with buying, selling and negotiating on our colleagues’ behalf.
1
3
242
The weird thing about building on Claude is that your product gets better while you sleep.😀
Claude Opus 4.7 launched! #anthropic #agent #opus4_7
Introducing Claude Opus 4.7, our most capable Opus model yet.
It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back.
You can hand off your hardest work with less supervision.
ALT Claude Opus 4.7 Benchmarks
2
217
join us on GitHub github.com/synctxai/synctx.g…
1
3
169