Web3 agent use case
LLM API for DAO and Web3 Community Bots
DAO and Web3 community bots answer questions, summarize proposals, and brief contributors inside Telegram and Discord. Router One is the LLM API behind those bots: one OpenAI-compatible endpoint, smart routing, a budget-capped API key per bot, and request traces — with USDT/USDC top-ups for teams that run treasury on stablecoins.
- Use case
- DAO and community assistant bots
- Router One role
- LLM inference gateway
- Channels
- Telegram, Discord, forums, dashboards
- Billing
- USDT/USDC top-ups, USD wallet
What DAO and community bots do
Proposal summaries
Condense long governance proposals and forum threads into a short brief for voters.
Governance briefings
Recap voting outcomes, quorum status, and treasury updates on a schedule.
Community Q&A
Answer recurring questions about docs, contracts, and processes inside Telegram or Discord.
Wallet activity recaps
Turn indexed treasury or wallet movements into human-readable updates for the community.
How the bot connects to Router One
Router One is the model layer. Your bot framework owns the Telegram or Discord connection and any chain data lookups.
One OpenAI-compatible endpoint
Point the bot at https://api.router.one/v1 and keep your existing SDK.
Your data sources
Use a snapshot, forum API, or indexer for proposal and chain data — Router One does not fetch it.
Smart routing and fallback
Route to GPT, Claude, Gemini, or DeepSeek, with automatic fallback when an upstream degrades.
Delivery in-channel
The bot posts the model response back into the Telegram group or Discord channel.
Budget control for always-on bots
One API key per bot
Issue a separate key for each bot, team, or environment so spend is isolated.
Hard budget caps
Set a spending limit so a busy channel or retry loop cannot run up an unbounded bill.
Request traces
Every reply records model, tokens, latency, and cost for reconciliation.
No advice, no execution
Build bots for briefings and Q&A. Router One returns model output and does not send transactions.
Community bot call pattern
Your bot receives a message, optionally fetches context, then calls Router One for the reply.
// 1. Bot receives a Telegram or Discord message
const question = message.text;
// 2. Ask a model through Router One
const reply = await client.chat.completions.create({
model: "auto",
messages: [{ role: "user", content: question }],
});
// 3. Post the reply back to the channelFAQ
- Can I pay for community bots with USDT or USDC?
- Yes. Stablecoin top-ups credit a USD wallet balance that every bot key spends per token.
- Can each bot have its own budget?
- Yes. Create a separate API key per bot with a hard budget cap so one channel cannot overspend.
- Does Router One connect to Telegram or Discord?
- No. Your bot framework owns the channel connection. Router One is the LLM API the bot calls.
- Does Router One fetch proposal or chain data?
- No. Use a snapshot, forum API, or indexer for that data. Router One reasons over the context you provide.
- Does Router One store bot messages?
- No. Request and response bodies are not retained; only metadata for billing and observability is logged.
- Can the bot give financial or voting advice?
- Router One is infrastructure. Build bots for summaries and Q&A, and keep disclaimers in your product.