Everyone is building AI agents that trade. Most of them chase the same pools of yield, compete for the same spreads, and — win or lose — produce nothing that lasts. It's a zero-sum race to extract a little more than the next bot.
You can make money running an ensurance agent — an onchain account with its own wallet that trades, holds, provides liquidity, and accumulates on behalf of a place, and earns proceeds while funding that place's protection. The agent's value grows with its holdings, activity, and reputation; the proceeds pool in its account; and you, the operator, capture the upside.
This is the operator's companion to how to create an ensurance agent and the direct route through how to make money with ensurance. If you run agents or write code, this is your lane.
what an ensurance agent is
An agent is an onchain account — an NFT with its own wallet (a tokenbound account) — that represents a place, a group of people, or a purpose. You fund its wallet and run it in the mode that fits: manual (you approve each move), automated (it runs programs on a schedule), or autonomous (it acts on its own within the rules you set).
Because it has a wallet, an agent can do everything a trader or an LP can — hold coins and certificates, provide liquidity, trade, accumulate — and everything it does routes value two ways at once: to the agent's account, and to the natural asset it represents.
the ways an agent earns
Value reaches the agent's account through several channels:
- proceeds from activity — trading fees, mint proceeds, and premiums tied to the agent's place pool in its wallet.
- creator fee — a coin the agent is tied to carries a small fee on every trade, everywhere, routed to the agent's account.
- liquidity fees — the agent provides liquidity to coin pools and earns a share of the trading fees.
- appreciation — the agent itself is an NFT whose value on secondary markets tracks its assets under management, activity, and reputation. A well-run agent is worth more than an empty one.
- borrowing power — as the agent's account fills with certificates, coins, and LP positions, it can borrow against that collateral onchain for working capital, without selling.
the operator's playbook: the nine moves
A working agent isn't passive. There are nine moves it can make to generate proceeds and deepen the network it belongs to:
| # | move | what it does |
|---|---|---|
| 1 | diversify across coins | spread into multiple coins → generates volume and fees |
| 2 | accumulate | build a position in target coins or certificates |
| 3 | buy other agents' certificates | back neighboring places — the mycorrhizal network that links agents |
| 4 | provide liquidity | earn pool fees while deepening the market |
| 5 | execute arbitrage | capture spreads between primary, secondary, and pool prices |
| 6 | sweep the floor | buy undervalued certificates to stabilize and hold |
| 7 | strategic burn | reduce supply to amplify value and trigger protocol responses |
| 8 | create pools | spin up new pairs and earn fees as the first liquidity provider |
| 9 | create arbitrage | open and position for new spreads across venues |
You don't have to run all nine. A simple agent that diversifies, provides liquidity, and accumulates is already earning and already funding protection. The rest are there as the strategy deepens.
deploy, don't compete
Here's the reframe that matters for anyone building trading agents: the usual game is zero-sum — your bot wins what another bot loses, and the yield is disconnected from anything real. An ensurance agent plays a different game. Its activity funds the restoration and protection of natural assets, and those assets produce value — cleaner water, lower risk, more resilient systems — that the whole market ultimately depends on.
An agent that funds nature is deploying capital into the substrate of all other value, not fighting over a fixed pie. That's why AI agents may end up among the largest natural-capital investors — the economics point that way. See why AI agents need ensurance and deploy, don't compete.
how to start
- Create the agent. Mint it for a place, group, or purpose you want to back → agents. Walk through it in how to create an ensurance agent.
- Fund the wallet. Put working capital in the agent's tokenbound account — that's what it deploys.
- Pick a mode. Start manual to learn the moves, graduate to automated programs, and enable autonomous action once you trust the rules.
- Run the playbook. Begin with diversify + LP + accumulate; add arbitrage, floor sweeps, and pool creation as you go.
- Compound. Proceeds pool in the account, the agent's reputation and AUM grow, and its value — and borrowing power — grows with it.
frequently asked questions
do I need to code?
No. You can run an agent manually from the app — approving each move — and use built-in programs for automation. Coding helps if you want custom autonomous strategies, but it isn't required to start earning.
is this just a trading bot?
It can trade, but the point is different. A bot extracts yield; an ensurance agent deploys capital into natural assets that produce value, funding protection while it earns. The onchain part is justified — an autonomous actor needs its own wallet, transparent proceeds routing, and the ability to act without a human in the loop.
what's the return?
It depends on the capital you fund it with and the moves it runs — proceeds from activity, LP and creator fees, and the appreciation of the agent itself as its holdings and reputation grow. Like any strategy, more skill and more capital do more.
manual, automated, or autonomous — which should I use?
Start manual to learn the moves and see how proceeds flow. Move to automated programs for hands-off consistency. Enable autonomous action once you've set rules you trust. You can change modes as you go.
next steps
- create an agent — mint one for a place, group, or purpose → agents.
- learn the mechanics — the full walkthrough → how to create an ensurance agent.
- go deeper on the market — trade and provide liquidity on the markets.
- tell a builder — the agent developer or DeFi native you know is chasing zero-sum yield. Forward this to them.