Router One

Web3 agent use case

Onchain Analyst Agent with LLM API and Stablecoin Billing

An onchain analyst agent turns raw wallet, protocol, and governance data into readable summaries. Router One is the LLM inference layer behind that agent: one OpenAI-compatible endpoint, smart routing, per-key budget caps, and request traces — with USDT/USDC top-ups for teams that fund AI usage from a stablecoin treasury. Your data provider supplies the chain data; Router One reasons over it.

Use case
Onchain analyst / research agent
Router One role
LLM inference gateway
Chain data
Your own indexer or data API
Billing
USDT/USDC top-ups, USD wallet

How an onchain analyst agent fits together

Router One does not fetch chain data. Pair it with a blockchain data provider — Moralis, The Graph, Dune, Alchemy, QuickNode, Etherscan, or your own indexer — and Router One handles the model call over the normalized data.

Data layer

Your indexer or data API returns normalized wallet activity, token flows, and protocol metrics.

Agent layer

Your service or framework decides what to fetch and when to call a model.

LLM layer

Router One routes the prompt to a suitable model, applies budget caps, and records a per-request trace.

Delivery layer

Summaries reach a dashboard, report, Telegram channel, or internal tool.

What an onchain analyst agent can summarize

Wallet activity

Explain transfers, swaps, approvals, and balance changes for a watched address in plain language.

Protocol metrics

Turn TVL, volume, fee, and emission data into a short read for contributors or investors.

Governance events

Summarize new proposals, voting outcomes, and treasury movements from your indexed data.

Risk signals

Describe unusual flows, large unlocks, or concentration changes as research notes — not trading calls.

Cost control and boundaries

Per-agent budget caps

Give the analyst agent its own API key with a hard spending limit before it runs on a schedule.

Request-level traces

Every model call records model, tokens, latency, status, and cost so agent spend stays visible.

No chain data or execution

Router One does not index chains, hold keys, or send transactions. It returns model output only.

Research, not advice

Frame outputs as research summaries. Your product owns disclaimers and any regulated workflow.

Analyst agent call pattern

Fetch chain context from your data provider, then send the normalized context through Router One for analysis.

onchain-analyst.ts
// 1. Pull onchain data from your provider
const activity = await getWalletActivity(address);


// 2. Summarize it through Router One
const report = await client.chat.completions.create({
  model: "auto",
  messages: [{ role: "user", content: summarize(activity) }],
});


// 3. Deliver to a dashboard, report, or channel

FAQ

Does Router One fetch onchain data?
No. Router One is the LLM gateway. A blockchain data provider or your own indexer supplies the data; Router One reasons over it.
Can I pay for the analyst agent with stablecoins?
Yes. Top up with USDT or USDC on supported networks; the value credits a USD wallet balance that the agent spends per token.
Can the agent have its own budget?
Yes. Issue a separate API key per agent with a hard budget cap so a scheduled run cannot overspend.
Does Router One store the prompts the agent sends?
No. Router One does not retain request or response bodies. It logs metadata for billing, routing, and observability.
Can an onchain analyst agent give trading advice?
Router One is infrastructure and does not generate advice. Build the agent for research summaries and keep user-facing disclaimers in your product.
Which models can the agent use?
Call GPT, Claude, Gemini, DeepSeek, Mistral, and Llama through one OpenAI-compatible endpoint, or let smart routing pick.

Related