Pinned Tweet
How to build an AI Agent that executes your strategy on @chance_ ?
In this video I break down exactly how I built mine, step by step.
Results so far: turned $150 into $159 in 8 hours. At this rate I overtake Elon by Christmas. Reality will probably have other plans, but the engineering is solid.
Try it completely free at ai.chance.cc (Use Code: AI)
Introducing Chance AI: the AI execution layer for Prediction Markets
Build your strategy, define your rules, and deploy an agent that trades for you on @Polymarket!
Instead of relying on generic bots or copy trading, you define everything yourself.
ai.chance.cc ← If you’ve ever felt like your ideas deserved better execution, this is where that changes.
In love with the new @Chance_ Harness UI!
Test it yourself: harness.chance.cc
Will you be there? 🫵
🇸🇬 Chance is heading to Singapore for @TOKEN2049!
Our team will be on the ground and would love to meet founders, investors, and builders!
See you in Singapore!
Agents are creative.
With real money, that is also the risk.
That's why we're building @Chance_
We didn't find a problem in a market map. We ran into it at @Chance_
We have an app with ~5,000 users moving real money. And we kept watching agents do things nobody asked them to do. Right market, wrong side. Right idea, 10x the size. Small mistakes, real losses.
Wallets couldn't catch it, because a wallet only knows you signed - not what you wanted.
So we built the thing that was missing.
One chart for why we're building @Chance_
The money AI agents move is climbing fast. The number of independent checks on what those agents actually do? Flat. At basically zero.
Hot take: "just add a spend limit" is not agent safety.
A spend cap sees a number. It can't tell an on-mandate $500 trade from a prompt-injected $500 one. It can't tell the right market from the wrong market. It stops big mistakes, not wrong ones, and wrong ones are most of them.
Real agent safety means checking the action against what the user actually asked for. Not the amount. The intent.
That's a harder problem. It's also the one worth solving. We solve it at @Chance_
The agent economy is going to need more than just smarter AI.
It needs infrastructure that makes sure agents can act without giving up control of the money they manage.
@Chance_ on @Starknet lets agents act while users keep control of the rules and funds.
Chance, now live on Starknet.
AI agents are increasingly being trusted to move money. Chance prevents agents from hallucinating with your money.
Before an agent transaction executes, Chance simulates the exact action and checks it against the user’s mandate. If it does not match, it is blocked before funds move.
On Starknet, Chance uses programmable escrow wallets. Agents can propose transactions, but funds only leave escrow when the proposed action satisfies the user’s mandate.
Additionally, Chance will use Starknet’s privacy framework, STRK20, to keep an agent’s mandate and settlement context private, while allowing receipts to be selectively disclosed when needed.
This is what Starknet enables for the agent economy: agents can act, but users keep control of the rules, the funds and what stays private.
Welcome, @Chance_
Chance, now live on Starknet.
AI agents are increasingly being trusted to move money. Chance prevents agents from hallucinating with your money.
Before an agent transaction executes, Chance simulates the exact action and checks it against the user’s mandate. If it does not match, it is blocked before funds move.
On Starknet, Chance uses programmable escrow wallets. Agents can propose transactions, but funds only leave escrow when the proposed action satisfies the user’s mandate.
Additionally, Chance will use Starknet’s privacy framework, STRK20, to keep an agent’s mandate and settlement context private, while allowing receipts to be selectively disclosed when needed.
This is what Starknet enables for the agent economy: agents can act, but users keep control of the rules, the funds and what stays private.
Welcome, @Chance_
Chance now supports @Starknet!
→ Intent verification for any agent transaction, before it executes
→ Settlement proofs anchored on Starknet
→ Native Starknet escrow wallets
AI agents don't need to be trusted. They need to be verified → harness.chance.cc
Your AI agent is one hallucination away from draining your account. We make sure that never happens.
Introducing Chance - the trust layer for AI agents that move money.
At its core: an independent AI judge that verifies every action against your actual intent, and blocks hallucinations, prompt injections and off-mandate trades before funds move.
Propose → verify → execute. Pass = executes, fail = nothing moves. Every verdict lands on-chain as a provable receipt - and every verdict makes both your agents and the judge sharper.
Battle-tested on our own prediction-market app. Now open to the ecosystem.
MCP live across @RobinhoodApp, @Polymarket, @HyperliquidX, @Uniswap, @OrderlyNetwork, @Lifiprotocol, @MyriadMarkets, @Trylimitless, @DeriveXYZ, @CoinbaseDev, @Agentcardai and @Dimes_fi.
Built alongside @Privy and @Alchemy
Any agent can move money. Only a trusted one should.
Docs: docs.chance.cc
Building agents that touch money? DMs open.
Everyone's building AI agents that can move money.
Almost nobody's asking the obvious question: what happens when the agent is wrong?
Because it will be. It'll buy the wrong market. Take a position 10x bigger than you asked for. Fall for a prompt injection. And it'll report back that everything went great.
Your wallet won't stop it. Your multisig won't either. They check that you signed - never whether it's what you actually meant.
That's the gap we're building Chance to close. An independent layer that checks every agent transaction against what the user actually asked for, and blocks the bad ones before the money moves.
The agent economy needs a safety layer. We're building it.
Your AI agent is one hallucination away from draining your account. We make sure that never happens.
Introducing Chance - the trust layer for AI agents that move money.
At its core: an independent AI judge that verifies every action against your actual intent, and blocks hallucinations, prompt injections and off-mandate trades before funds move.
Propose → verify → execute. Pass = executes, fail = nothing moves. Every verdict lands on-chain as a provable receipt - and every verdict makes both your agents and the judge sharper.
Battle-tested on our own prediction-market app. Now open to the ecosystem.
MCP live across @RobinhoodApp, @Polymarket, @HyperliquidX, @Uniswap, @OrderlyNetwork, @Lifiprotocol, @MyriadMarkets, @Trylimitless, @DeriveXYZ, @CoinbaseDev, @Agentcardai and @Dimes_fi.
Built alongside @Privy and @Alchemy
Any agent can move money. Only a trusted one should.
Docs: docs.chance.cc
Building agents that touch money? DMs open.
Hot take: most "AI trading agents" are just bots.
Most aren't really AI. They're rule engines with a language model bolted on for marketing copy.
Real agent behavior means: the system reads the market, reads the news, forms a view, checks it against your constraints, decides whether to act, and explains its reasoning, in a loop, autonomously.
That's what we built at @Chance_
As a founder, the most exciting moments are when the industry starts independently arriving at the same conclusion you built around from day one.
AI-native trading isn’t a feature.
It’s the next interface layer for markets.
This revolution is already underway.
Today @RobinhoodApp announced AI agents can trade on their platform.
It confirms exactly what we believed when we started building @Chance_
The future of trading isn't you watching a screen. It's an agent executing your strategy.
The revolution is happening.
.@HyperliquidX was one of the first communities I truly felt a part of.
I met so many brilliant, genuine people there. Builders, traders, researchers, anons with insane conviction. It gave me a front-row seat to what a world-class product and ecosystem should look like when it’s built with obsession and clarity.
A lot of what we’re building today at @Chance_ was shaped by those experiences, directly or indirectly.
So integrating Hyperliquid HIP-4 markets into Chance feels personal to me. It feels like bringing a piece of my own journey into the product we’re creating.
Can’t wait to integrate HL to AI-powered agentic trading strategies very soon.
Chance has integrated @HyperliquidX HIP-4 into the Terminal.
You can now trade outcome markets from Hyperliquid directly on @Chance_
Use code 'HYPE' to access the platform.
Agentic trading strategies on Hyperliquid - coming soon!
Introducing Chance AI: the AI execution layer for Prediction Markets
Build your strategy, define your rules, and deploy an agent that trades for you on @Polymarket!
Instead of relying on generic bots or copy trading, you define everything yourself.
ai.chance.cc ← If you’ve ever felt like your ideas deserved better execution, this is where that changes.
As CTO at @Chance_, I’m genuinely excited about what’s coming in the next few weeks.
We’ve been building a lot behind the scenes, rolling out new features and improvements, starting today with the @trylimitless integration.
What’s next? Stay tuned.

















