The best part of building early is having a place to show the rough edges.
We’re sharing the walkthroughs, the tests, the questions, and the build as it becomes real.
If that’s your thing, come into the chat:
https://t.co/2B1jLmjT3G
A security question we hear often: could OKO ever access funds it shouldn’t?
Your exchange API key permissions define what any trading tool can do.
For OKO, use trading only keys, keep withdrawals disabled, and add an IP allowlist wherever your exchange supports it.
Those are baseline controls, not a favor to us.
https://t.co/AJcO60xvgw
One thing we didn't want to build is a black box that just says "trust the algorithm."
OKO supports seven (7) plus AI models and lets you run them against each other on a live leaderboard, so you can actually see which one performs better instead of assuming.
gm
We’re back in the build: testing strategies, tightening the workflows, and making sure every AI driven decision has a clear trail behind it.
More to show soon, Drop EVM Wallets (Robinhood)
gm
We’re back in the build: testing strategies, tightening the workflows, and making sure every AI driven decision has a clear trail behind it.
More to show soon, Drop EVM Wallets (Robinhood)
Before OKO puts a strategy anywhere near real funds, it runs through Backtest Lab against historical data:
return, max drawdown, Sharpe ratio, win rate, trade by trade.
We'd rather you find out a strategy doesn't work in a backtest than in a live account.
OKO is built for the part most AI trading tools skip: the work before execution.
Live market context. Strategy design. Historical testing. Model debate. A decision trail you can actually inspect.
That is the stack. Almost ready.
@RobinhoodApp
OKO is open-source (AGPL-3.0) and self-hosted.
There's no OKO server holding your exchange keys.
Everything runs on your own machine, and the code is public if you want to check that claim yourself instead of taking our word for it
https://t.co/n2BZxKPL1M
We’re building a clearer way to work with AI trading: test the thesis, inspect the reasoning, and keep the decision trail visible.
More soon, Drop EVM wallets (Robinhood)
Open Strategy Studio
Coin Source / pick what to trade (static list, AI500 pool, or OI Top ranking)
Indicators / toggle EMA, MACD, RSI, ATR, Volume, Open Interest, Funding Rate
Risk Controls / set leverage limits, max positions, margin cap, position sizing
Custom Prompt / give the AI specific instructions
Save, then assign it to a Trader to go live.
Backtest it before you trust it with real funds.
OKO team / stay safe.
The interesting part of agent native finance is not simply letting an agent transact.
It is giving the agent clear boundaries and giving the user a way to understand what happened.
That is a direction worth building toward. @RobinhoodApp
The interesting part of agent native finance is not simply letting an agent transact.
It is giving the agent clear boundaries and giving the user a way to understand what happened.
That is a direction worth building toward. @RobinhoodApp
If you're experimenting with AI-driven trading, our honest advice is the same we'd give anyone: start small, backtest before you go live, and don't put in more than you're fine losing. OKO is a tool for that experimentation, not a guarantee of returns.
Stay safe ...
Access matters. So does understanding the systems people use to make decisions.
OKO is building AI assisted trading workflows where you can test a thesis, review the reasoning, and see the process not just the outcome.
One thing we didn't want to build: a black box that just says "trust the algorithm." OKO supports seven-plus AI models and lets you run them against each other on a live leaderboard, so you can actually see which one performs better instead of assuming.
- OKO team.
New day, same focus: making AI trading easier to inspect.
Build the strategy. Test the idea. Review the reasoning. Then decide what deserves a live run.
The OKO beta is open: https://t.co/n2BZxKQiRk
Check it out and let us hear your feedback
We are building OKO in public, and the useful conversations are happening before launch.
If you want the walkthroughs, beta updates, and the less polished parts of the build, join us here:
https://t.co/2B1jLmkqTe
New day, same focus: making AI trading easier to inspect.
Build the strategy. Test the idea. Review the reasoning. Then decide what deserves a live run.
The OKO beta is open: https://t.co/n2BZxKQiRk
Check it out and let us hear your feedback
Access matters. So does understanding the systems people use to make decisions.
OKO is building AI assisted trading workflows where you can test a thesis, review the reasoning, and see the process not just the outcome.
We are close enough to start showing the work, not just talking about it.
Walkthroughs will cover the parts that matter: market data, strategy building, testing, model perspectives, and the records behind each decision.
Launching soon on @RobinhoodApp
Beta testers, this is what we need from you: try the path from an idea to a strategy, push the workflow, find the confusing parts, and tell us what would make you trust or reject an AI assisted decision.
beta website: https://t.co/AJcO60xvgw
Tg for feedback: https://t.co/TXVDo1j81q
Tokenized markets are going to be 24/7, which means decision systems need to be more than fast. They need to be explainable when nobody is watching.
That is why @RobinhoodApp building for agents is interesting and why OKO is focused on the control plane around the agent.