The intelligence layer for onchain finance
A unified gateway to wallet intelligence and financial workflows on Robinhood Chain. Connect your applications and AI agents through MCP.
Built around open standards
One foundation for onchain intelligence
Wallets
Wallet intelligence, ready for your applications
Access supported balances, canonical asset metadata, and network state through a consistent interface.
Read native balances
Query ETH balances on Robinhood Chain using a public wallet address.
Discover supported assets
Start with ETH, WETH, and USDG using canonical network addresses.
Connect any MCP client
Give compatible AI assistants a structured, read-only view of the chain.
MCP tools
A direct interface between agents and the chain
Connect a compatible MCP client and discover tools for wallets, tokens, and network state.
Discover available tools
Use standard MCP discovery to see each tool, its inputs, and the data it returns.
Work with open standards
JSON-RPC, ERC-20, and MCP provide a familiar foundation for your integration.
Keep your wallet separate
Read public chain data without depositing funds or sharing a private key.
Agents
Financial workflows with inspectable results
Run a hosted portfolio inspection workflow powered by the same tools your MCP client can use.
A ready-to-run workflow
Give the portfolio workflow a public address. Envoy reads supported balances and returns a structured report.
Read-only by default
The gateway exposes data tools. It cannot sign transactions or move your funds.
Inspectable run results
Each workflow reports its status, the data it read, and a summary you can inspect.
Developers
Designed for your existing stack
Call the gateway from your app or connect an MCP client. Start with a single request.
A small, useful toolset
Read network state, ETH balances, and supported token balances through documented tools.
Structured responses
Receive machine-readable results ready for dashboards, assistants, and workflows.
One asset gateway
Use the same data layer from a browser playground or a compatible MCP client.
A shared foundation for onchain applications
Useful primitives for financial agents
Network state, native balances, supported token balances, and an asset directory — available as structured MCP tools.
Canonical assets. Explicit context.
Keep the network, contract address, and token decimals alongside each result. Avoid relying on a ticker alone.
Read the chain. Keep control.
Envoy queries public blockchain data. Wallet signing stays outside the gateway, under your control.
Built on an open financial network
Ethereum-compatible foundation
Build on Robinhood Chain with familiar Ethereum addresses, JSON-RPC calls, and ERC-20 contracts.
Public data, on demand
Request the data you need when your workflow needs it. No wallet connection is required for public balance lookups.
Crypto assets in context
Inspect ETH, WETH, and USDG balances with explicit chain and contract information.
Honest network status
Read the latest block and gas price directly from the configured RPC. Upstream failures appear as errors, not invented data.
Start with visibility. Build with control.
Asset Gateway and Agent Cloud, together in a developer preview: live chain reads, MCP tools, and a hosted portfolio workflow. The diagrams below illustrate application patterns.
Make every action explicit
Start with read-only tools. Keep signing and execution in your wallet. Add policies and approval flows before enabling any financial actions.
Context your agents can inspect
Return the chain ID, asset identity, and raw values with every balance result so applications can inspect the underlying data.
- Time
- 05:17am
- Containers
- 108
- GPU Utilization
- 94%
A focused developer preview
Run the portfolio workflow or call its tools over MCP. Custom agent deployment, scheduling, and financial execution are not enabled in this preview.
Use cases
“Give a portfolio assistant the public balance data it needs, without handing it the keys to your wallet.”
Latest updates
Launching a trillion-parameter model at Sakana AI
Introducing our MCP tools API
Spindle is an open-source API for training and fine-tuning open models — deploy a server with a single command and run any Tinker-compatible training code against it.
Open playground with LLM training on Modal
Reinforcement learning for LLMs is hard, so we built Modal Dojo — a library that gives you the building blocks, tuned defaults, and observability to get started with production-grade LLM training.