While the industry debates whether autonomous agents are hype or reality, a fully operational agent economy has been running in production for years. It's called the MEV supply chain -- and it's the most sophisticated example of autonomous software competing in a real market with real money at stake.
Searchers scan mempools for profitable transaction orderings. Builders assemble optimal blocks. Relayers connect builders to validators. Every participant in this chain is autonomous software making high-stakes economic decisions in milliseconds, with no human in the loop. This isn't a proof of concept. Cumulative MEV extraction on Ethereum has reached billions of dollars since 2020. Flashbots Protect alone routes a significant share of Ethereum transactions, shielding users from sandwich attacks while enabling a structured marketplace for block space. The strategies are diverse and increasingly sophisticated: sandwich attacks on DEX trades, just-in-time liquidity provision on Uniswap v3, backrunning of large oracle updates, and cross-DEX arbitrage that keeps prices aligned across venues.
Flashbots and the Architecture of Agent Markets
Flashbots didn't just mitigate MEV -- they created the infrastructure for agent-to-agent commerce. MEV-Boost established a marketplace where autonomous software entities bid for block space, negotiate inclusion, and settle payments. The protocol handles identity (builder registration), commerce (bid submission), and dispute resolution (relay verification) -- all without human intervention. This is exactly the pattern we expect to see in every vertical where agents operate.
The MEV supply chain solved the coordination problem first because the economic incentives were immediate and enormous. But the architectural patterns -- permissionless participation, automated bidding, cryptographic settlement -- will generalize to any domain where autonomous software transacts at scale.
The formalization runs deeper than most people realize. Proposer-Builder Separation (PBS) isn't just a technical optimization -- it's the codification of distinct agent roles within a market. Proposers (validators) delegate block construction to specialized builders, who compete on execution quality. This separation of concerns mirrors how mature economies develop: generalists give way to specialists, and market infrastructure emerges to coordinate between them. Within the builder role alone, further specialization is emerging -- some builders optimize for DEX arbitrage inclusion, others for CEX-DEX flow, and others for long-tail transaction bundles. Specialization begets more specialization, and infrastructure is needed at every layer.
MEV-Share takes this further by redistributing extracted value back to the users whose transactions created the opportunity. This is agent cooperation, not just extraction -- a protocol-level mechanism where searchers share profits with order flow originators. It's an early example of what cooperative agent economics might look like: agents competing on execution quality while sharing value with the ecosystem that generates their opportunities.
The relay landscape itself has become a competitive market: Flashbots, bloXroute, Ultra Sound Relay, and Agnostic Relay each offer different trust assumptions and performance characteristics, and builders route through multiple relays simultaneously to maximize inclusion probability. Even the relay layer -- which is infrastructure for infrastructure -- has become a multi-player competitive market driven by autonomous participants.
The Supply Chain Keeps Growing
The next evolution is already underway, and it's expanding the supply chain rather than consolidating it. Intent-based protocols like CoW Protocol and UniswapX represent a fundamental shift: instead of users submitting specific transactions, they express intents ("swap X for Y at the best price") and a network of solvers competes to fulfill them. Solvers are a new agent role -- they don't just search mempools for existing opportunities, they actively construct optimal execution paths across liquidity sources, chains, and time horizons. The solver market alone is becoming a multi-billion dollar opportunity as more protocols adopt intent-based architectures.
Cross-chain MEV is extending this supply chain beyond Ethereum. SUAVE from Flashbots aims to create a unified execution environment for cross-chain intents. Fastlane and Skip Protocol are building MEV infrastructure for Cosmos ecosystem chains. The MEV supply chain isn't consolidating around a single chain -- it's expanding into a multi-chain agent economy where opportunities span bridges, rollups, and alternative L1s. Each new venue adds complexity, and complexity is where specialized agents thrive.
The key insight is that the supply chain is growing, not consolidating. New agent roles keep emerging. Solvers didn't exist three years ago -- now they're essential infrastructure. Block builders were irrelevant before the Merge -- now they process every Ethereum transaction. We should expect entirely new agent roles to emerge as intent-based architectures mature and cross-chain complexity increases. The teams building infrastructure for roles that don't exist yet -- but inevitably will -- are the ones we want to find early.
From MEV to Everything
The MEV supply chain proves that autonomous agents can sustain a complex, multi-party economy without centralized coordination. The implications extend far beyond block building. Every market where speed, information asymmetry, and automated execution create value is a candidate for the same architecture.
We're watching this pattern emerge in liquidation markets, cross-chain arbitrage, intent-based trading, and on-chain order flow. Each of these domains is developing its own supply chain of specialized agents -- and each one needs the same infrastructure primitives that MEV-Boost pioneered. Lending protocol liquidations on Aave and Compound already depend on searcher bots that compete for liquidation opportunities. DEX aggregation through 1inch and Paraswap relies on routing algorithms that are, functionally, agent-like solvers. The pattern is everywhere once you know what to look for.
The Investment Thesis
If you want to understand where the agent economy is going, study where it already exists. The MEV supply chain isn't an anomaly -- it's a preview. At Isoline, we invest in teams building the generalized infrastructure for agent markets: intent routing systems that translate agent goals into optimal execution, cross-chain execution layers that let agents operate across any chain, and solver infrastructure that powers the next generation of agent-to-agent marketplaces. The playbook that MEV pioneered is about to run everywhere.