Ymax is Agoric’s non-custodial, cross-chain capital management application. It is not a vault and does not pool user funds. Users retain ownership of capital deployed across protocols and chains, with every action visible prior to authorization and revocable at the permission layer. A single signature initiates complex, multi-step allocation workflows that would otherwise require multiple manual transactions across fragmented interfaces and protocols.
The coordination problem Ymax addresses extends beyond interface complexity. As automated execution tools become more common in DeFi, users increasingly face a trade-off between convenience and control: do everything manually, or hand over broad control (wallet keys, broad permissions), and trust the tool behaves as intended. There is no middle ground. Ymax addresses both dimensions of the coordination problem by combining multi-step orchestration with scoped execution. Users can automate complex, multi-step portfolio strategies across chains and protocols while defining exactly where capital can go and what actions are permitted. Those boundaries are enforced onchain rather than promised in a terms-of-service.
Agoric's object-capability (Ocap) security model provides the architectural foundation for this approach. Under this model, an agent authorized to optimize or rebalance within approved protocols cannot withdraw to an unapproved address or act outside its defined scope. The enforcement mechanism operates at the execution layer rather than relying on external policy or monitoring.
Early user research and beta testing point to a real and repeatable behavioral signal: when execution complexity is reduced, and workflow visibility is preserved, users become significantly more willing to keep capital actively allocated. The primary barrier to yield optimization does not appear to be a lack of awareness of opportunities, but rather the operational cost of acting on them.
Ymax's development roadmap moves from yield orchestration to the capital execution layer supporting user-configured automation and agent-driven capital allocation. Planned extensions include rule-based optimizations and rebalancing, hosted agent interfaces, and Bring Your Own Agent (BYOA) integrations. Initial experiments, including work with Sommelier, illustrate how external strategy systems may interact with orchestration infrastructure without requiring custody transfer.
Primer
Ymax is the first flagship application built on Agoric’s orchestration infrastructure. In its current form, it functions as an onchain command center for capital, allowing users to define portfolio-level allocation targets and authorize execution across multiple chains and protocols through a single approval flow. Ymax is building toward a trust layer for automated capital: it allows both humans and AI agents working on behalf of power users to manage capital efficiently onchain.
The distinguishing characteristic of Ymax is its underlying execution model, not solely interface simplification, though user experience improvements are material. Under this model, every action is transparent, non-custodial, and previewable. More importantly, when agents act on capital through Ymax, their authority is scoped, revocable, and enforced onchain, instead of by a third-party fintech like Stripe. The AI plans offchain. The smart contract enforces onchain. The user retains override rights at all times.
Ymax is built on Agoric, a Proof-of-Stake blockchain secured by CometBFT and integrated natively with Inter-Blockchain Communication (IBC) for multichain composability. Agoric's Hardened JavaScript environment and Endo framework mean smart contracts run in an isolated, tamper-resistant context with only the specific, fine-grained authority they are granted. This environment supports asynchronous, multi-step workflows spanning multiple chains. Agoric smart contracts also support execution on non-IBC chains, including Ethereum via third-party SDKs.
As of March 2026, Ymax is in early access beta, supporting 13 stablecoin yield opportunities, including RWA and lending, across Aave, Compound, and Morpho on Ethereum, Arbitrum, Avalanche, Base, and Optimism. All currently supported pools accept USDC, with stated APYs ranging from 1.62% to 32.50%.
For a detailed overview of Ymax, refer to our prior Pulse report.
As DeFi has expanded across chains and protocols, the primary constraint on capital efficiency has shifted from identifying yield opportunities to maintaining exposure to them. Information about where yield is highest is widely available. The problem is that executing reallocations across chains requires users to manually coordinate bridging, approvals, deposits, and rebalancing across multiple environments, and that operational overhead is high enough that even sophisticated users often tolerate suboptimal allocations because the cost of continuously maintaining portfolio positioning outweighs the incremental yield improvement.
A crosschain yield analysis the Ymax team conducted with Sommelier illustrates this dynamic. Across a 26-month backtest covering Ethereum, Base, and Arbitrum, no single chain consistently offered the highest USDC lending yield. An allocation strategy confined to Ethereum generated an annualized return of 9.23%. The highest performing single-chain environment, Arbitrum, generated 12.86%. A cross-chain strategy rebalancing weekly achieved 12.69% after accounting for bridge costs. These results suggest that maintaining exposure across chains can materially improve realized yield without requiring predictive assumptions about which ecosystem will dominate over time.
As incentive programs and liquidity migration accelerate across networks, the gap between theoretical yield availability and realized yield is expected to widen. The limiting factor increasingly becomes execution coordination rather than opportunity discovery.
Existing abstractions partially address this problem. Vaults simplify access to predefined strategies but typically require pooled capital and reduce transparency into underlying allocation logic. Dashboards improve visibility but do not execute transactions. Transaction aggregators optimize individual routing decisions but are not designed for continuous portfolio-level maintenance across chains.
Yield orchestration represents an emerging infrastructure layer designed to address this coordination problem. Rather than generating yield itself, orchestration systems allow users to express allocation intent at the portfolio level while software coordinates the multi-step execution required to implement that allocation across chains and protocols.
Ymax provides an implementation of this design pattern built on an execution layer capable of enforcing constrained delegation of authority. This design choice becomes increasingly relevant as automated capital allocation tools become more prevalent. Automation introduces efficiency benefits, but also raises requirements for verifiable constraints on how capital may be deployed.
Ymax Overview
In its current form, Ymax users connect a wallet, fund with supported stablecoin assets, specify target allocations and execution constraints, review the proposed workflow, and approve it with a single signature. Once the single signature is executed, the required transactions are coordinated across supported protocols and chains without further manual input.
Rather than executing as a single atomic transaction, the workflow unfolds as a sequence of dependent steps. The system initiates each action, waits for confirmation or settlement, and proceeds as execution progresses. This can include reallocating capital across lending protocols, transferring assets between chains, and updating portfolio positions until the target allocation is reached.
This execution model does not require Ymax to take custody of user assets. Capital remains in user-controlled wallets and protocol positions throughout the workflow, and all actions require explicit approval before execution begins. The high-level intent of the portfolio change is presented, with steps then generated dynamically after signing to satisfy this high-level intent. Execution can then be monitored as it progresses.
'By using their own chain's support for very readable messages, Ymax was able to encode the most readable signature challenges into a MetaMask ERC-712 signature that I've seen yet, and in web3, readability is a requirement for safety.'
The Architectural Differentiation
Agoric’s execution model combines several properties relevant to constrained automation.
Hardened JavaScript + Endo
The execution environment is locked down. Smart contracts or AI generated code run in an isolated, tamper-resistant context where objects cannot be unexpectedly mutated. This is a language-level property, not a policy. MetaMask chose this architecture to safely isolate Snaps for 30 million users. Salesforce uses a co-developed SES implementation for 5 million developers. This is mature, externally validated technology, not a roadmap item.
Object-Capability (Ocap) Security
Agents and contracts only wield the authority that they are explicitly granted. They do not operate with the user or service operator’s authority. An agent authorized to rebalance within approved protocols cannot, by construction, withdraw to an unapproved address, bridge to unapproved destinations, or act outside its defined scope. Competitors using API keys or broad wallet permissions cannot make this guarantee. As Agoric CEO Dean Tribble notes, the investment agent might put capital into a low-yield fund, but it cannot send it to another party or burn it.
Confined Agent Execution
Agents operate on user assets via the Ymax smart contract. The agent does not have user keys, act with the user’s direct authority, have access to bridges, etc. When an AI agent acts on capital through Ymax, its authority is scoped, revocable, and enforced onchain. It can only express the high-level intents for the portfolio, such as, move $1,000 USDC from Aave on Base to Morpho Gauntlet RWA on Ethereum mainnet. The AI planning layer operates offchain, providing strategy and optimization. The chain enforces what is actually executed. The user, or the principal who set up the agent's authority, retains the ability to revoke, override, or modify the agent's scope at any time. This also enables dynamic delegation: a user can grant an agent read access to a portfolio and delegate a subsidiary agent to perform analysis, passing only that read-only access, nothing more.
The competitive landscape illustrates the gap this fills. Wallet-layer tools (MetaMask Smart Accounts, session keys) grant permissions at the transaction level but do not solve portfolio allocation orchestration and do not provide confinement at the strategy layer. Intent-based networks operate at execution routing, not portfolio strategy. Autonomous agent platforms trend toward full autonomy with custody risk, or require agent developers to build their own execution rails with no safety guarantees. The gap no solution currently occupies is a portfolio strategy layer where agents optimize allocations within user-defined guardrails, execution is non-custodial and auditable onchain, and confinement is enforced by smart contract, rather than trust in a service to uphold its terms of service after granting full authority. Agoric's object-capability security model makes this possible in a way that wallet-layer or API-key approaches fundamentally cannot replicate.
Early Evidence of Product Relevance
Agoric’s evidence base for Ymax consists of team-reported user research and early product testing. According to the team, this included 149 discussions with DeFi power users at ETHDenver, 13 recorded private beta interviews, and direct testing with 30 experienced users during the private beta and early access phases. While this does not constitute independent validation or broad market demand, it is sufficient to identify repeated patterns among likely early adopters.
Reducing execution burden changes user behavior
The most consistent finding across beta testing is that users were generally already aware of yield opportunities but delayed acting on them because execution required multiple manual steps. During testing, users reportedly became more willing to rebalance or reallocate when those workflows were compressed into a single process. One tester described noticing their capital was no longer in the highest yield opportunity and feeling the urge to immediately reallocate, not because they had to, but because the process finally felt simple enough. This reaction appeared repeatedly. Once users understood that Ymax could handle multi-step execution, they began thinking in terms of continuous allocation rather than isolated transactions. The implication is that some opportunities go unacted upon not because they are unknown, but because acting on them is operationally costly.
Stablecoin allocation is the validated entry point
Early testing concentrated primarily around USDC deployment across supported lending markets. Stablecoin strategies offer a particularly useful initial wedge for an orchestration system because they allow users to evaluate execution quality independently of directional market exposure. Performance can be assessed without the confounding variable of price movement. Usage centered on repeated deposit, withdrawal, and reallocation flows across protocols and chains rather than complex portfolio constructions. This is why Ymax intends to build execution trust through the most legible and isolatable use case first, before expanding scope.
Institutional signal from the beta period
A notable data point from early access is inbound interest from institutional participants exploring Ymax for stablecoin treasury management. Protocol treasuries, crypto funds, and onchain treasury operators managing significant USDC positions face the same structural problem as individual power users. Yield varies meaningfully across chains and protocols, and maintaining an optimized position requires either constant manual effort or accepting suboptimal allocation. Ymax's non-custodial, auditable, one-signature execution model maps directly onto institutional treasury requirements with no custody risk, full transparency, and the ability to move capital across protocols and chains without requiring multiple signers or manual bridging workflows. While still early, this points to a broader addressable market beyond individual DeFi power users
Strategic Direction and Product Vision
Ymax's near-term roadmap is shaped by two converging forces: what early users are asking for, and where the broader DeFi ecosystem is heading, both of which point to the same infrastructure.
On the user side, beta feedback consistently surfaces the same unmet need that users don't want another vault or dashboard. They want capital that stays actively deployed without constant manual attention. Users who have experienced one-signature execution naturally ask whether this can happen automatically when yield conditions shift. That question is not primarily a UX question. It is a trust question. And the answer, for almost every automation tool today, is that it cannot, because you would have to give away full control of your funds, full control of your keys, or both.
DeFi is heading towards adoption of AI agents for research, analytics, opportunities discovery and even autonomous economic activity. The missing infrastructure layer is one that will allow agents to deploy capital without custody, within strictly defined and enforced boundaries, a layer that Ymax will provide.
Near-term: 60-second onboarding and automated stablecoin yield
Ymax’s immediate product priority is removing remaining friction from the core user journey. The target is a user going from zero to a fully optimized, crosschain stablecoin yield portfolio in under 60 seconds, with no manual bridging, no chain selection, and no protocol research required. This also includes near-term automation capability that enables users to set allocation rules that execute automatically when yield conditions shift, without requiring a manual signature for each rebalance.
Medium-term: Three lanes to confined agent execution
The core of Ymax's medium-term strategy is enabling AI agents to act on capital safely via a go-to-market approach with three complementary lanes.
Lane 1 — Bring Your Own Agent (BYOA)
This is for users who already have an agent (Claude or similar instance), a custom script, a quant tool, etc., and want to give it delegated access to their Ymax portfolio. The agent can reallocate within approved protocols, respond to yield shifts, and execute strategies, but cannot withdraw funds, bridge to unapproved destinations, or act outside the scope the user has defined. The Sommelier integration is the first live proof point of this model. Sommelier connected its risk engine to Ymax to apply strategy to stablecoin deploymentwithout taking custody. Importantly, the delegation and permission infrastructure required for this lane is the same infrastructure Ymax already needs for the hosted agent lane. This lane requires relatively little additional development and gives Ymax a credible, differentiated story in agent developer communities before the hosted agent is fully built.
Lane 2 — Hosted Agent (Max)
This is for users who want an agent but don't want to build one. Ymax provides Max, a pre-built, hosted agent that understands the user's portfolio history, investment preferences, and strategy, and can execute on their behalf within user-defined guardrails. The initial version presents a small number of predefined strategies (maximize yield, rebalance weekly to top-performing chain, stay within a specific protocol) with a custom option for free-form instructions. Over time, Ymax intends for Max to grow into a strategy-aware agent that maintains context across sessions.
Lane 3 — Orchestrated Execution APIs
Making available Ymax's orchestration infrastructure as an API layer for protocols, strategy developers, and institutional operators to build on top of.
Agent-layer architecture
Delegating to agents also simplifies introduction of new asset types and protocols. Building a clear and usable user interface for a new financial arrangement can be challenging, as each integration requires designing for edge cases, withdrawal queues, and unbonding periods that are hard to communicate clearly to human users. Agents handle this natively. An agent can be told that a withdrawal from a given protocol takes an hour and requires a follow-up action. It doesn't need a UI built around that complexity. The agent layer lowers the barrier to expanding the Ymax universe of supported protocols and assets.
Longer arc: agents as economic participants
Ymax’s longer-term thesis goes further than agent-assisted portfolio management. The direction is toward a world where AI agents are not just curators helping humans allocate capital, but rather participants in investment services, deploying capital on behalf of principals or operating as economic actors in their own right. For this world to function safely, the infrastructure layer connecting agents to financial execution must provide the right level of confinement (enough authority to act, but narrow enough to prevent misuse). It is under this thesis that Ymax has been designed, as the properties that make Ymax valuable for human users today are the properties required to make agent-driven capital deployment trustworthy at scale.
Closing Summary
Ymax addresses the specific and real coordination problem of keeping capital deployed efficiently as opportunities shift across protocols and chains. The stablecoin-focused initial use case, portfolio-level execution, and non-custodial workflow design reflect correct sequencing for build execution trust through the most legible use case first.
The more significant thesis, however, is structural. Ymax is positioned at the intersection of two converging forces: the fragmentation of DeFi yield across chains, and the emergence of AI agents as autonomous economic actors. The primary constraint in DeFi capital management is no longer identifying opportunities. It is maintaining exposure to them, and doing so in a way that does not require handing over custody or control.
Agoric's object-capability security model makes it possible to enforce agent boundaries onchain, by construction, in a way that wallet-layer or API-key approaches fundamentally cannot. That architecture is in production today.
Early Ymax team-reported research and beta testing suggest demand among a validated subset of power users. The open questions are the right ones, including whether users continue to rely on the Ymax system rather than reverting to manual workflows, whether execution remains transparent and predictable under broader usage, and whether the initial stablecoin scope expands successfully into agent delegation and more complex allocation strategies. The product is being built against a real constraint set, on infrastructure that competitors cannot easily replicate. Its broader relevance will depend on execution quality and user retention, but the architectural foundation for the category-defining position is already in place.
This report was commissioned by Agoric Systems Operating Company. All content was produced independently by the author(s) and does not necessarily reflect the opinions of Messari, Inc. or the organization that requested the report. The commissioning organization may have input on the content of the report, but Messari maintains editorial control over the final report to retain data accuracy and objectivity. Author(s) may hold cryptocurrencies named in this report. This report is meant for informational purposes only. It is not meant to serve as investment advice. You should conduct your own research and consult an independent financial, tax, or legal advisor before making any investment decisions. Past performance of any asset is not indicative of future results. Please see our Terms of Service for more information.
No part of this report may be (a) copied, photocopied, duplicated in any form by any means or (b) redistributed without the prior written consent of Messari®.
Matt is a Research Manager at Messari for the Protocol Reporting team. A generalist at heart, who's curious about anything and everything, and ultimately, on an adventure to find out what's true. He was an investigative reporter and multifamily/senior housing development associate before joining Messari in 2022.
Matt is a Research Manager at Messari for the Protocol Reporting team. A generalist at heart, who's curious about anything and everything, and ultimately, on an adventure to find out what's true. He was an investigative reporter and multifamily/senior housing development associate before joining Messari in 2022.
