PropAMMs are live on Ethereum! pAMM liquidity is now available in every Titan block.
This is a major step toward broader Application-Controlled Execution (ACE) on Ethereum.
Pearl mainnet is live.
Since 9am UTC, April 27, 2026 blocks are being mined.
You can now run Llama70b (and in a few days, Gemma4) with Pearl’s 2-for-1 GPU kernels via our vLLM plugin. Every inference cycle simultaneously mines ¶Pearl -- All the properties of Bitcoin's Proof-of-Work, but secured as a byproduct of native AI workloads = arbitrary matrix multiplications.
Pearl is changing the unit economics of training and inference. A parallel revenue stream for every GPU cycle powering AI systems.
AI’s native currency.
https://t.co/HKqsuJKGfL
https://t.co/JjMYXi6S8x
What you're watching: a swap on Gnosis mainnet routed through L2 liquidity on Surge. Bridge, swap, settlement, all in a single L1 transaction.
ZK proven by @ziskvm
https://t.co/uU4PJjBQlE
For a vault with 4+ protocol dependencies - lending market, perp venue, RWA issuer, bridge: does the annualized exploit risk leave anything on the table over T-bills?
Nobody is calculating the only metric that matters in DeFi vaults: all-time yield over T-bill, net of all money ever stolen from the protocols inside them.
Is it even positive?
Just got back from Cannes. A lot of the institutional yield teams has pivoted to DeFi vaults.
Most strategies under the hood are RWAs (essentially a token issuance teams), lending markets, perp DEXs vs spot arb and exotic DeFi primitives. These are historically the most targeted attack surfaces in crypto. Both logically and empirically.
One blind spot worth flagging -
Simulation-based overquote catches dishonest routing, but not frontrunning. A monetization strategy (e.g. aggregator frontrunning their own flow, based on the trader profile) only activates on real submissions, so an aggregator could look clean on your metric while still extracting value in practice.
Ground truth requires comparing quotes to actual onchain output. Hard to do at scale, but it's the only way to fully capture this.
The "Flow Owner Era" thesis isn't about one solution winning.
It's about protocols recognizing their own valuable order flow, and deploying the right tools for each arbitrage regime they generate.
There's a distinction in MEV that doesn't get enough attention:
Not all arbitrage opportunities are created equal.
Understanding the difference between marginal and contested arbitrage explains why no single mechanism captures all MEV.
A thread 🧵
This is what we built Reflex to solve.
Not to replace pricing mechanisms, but to complete the picture. PFDA handles the margins. Reflex handles the contested executions.
Together, protocols capture MEV across both regimes.