Binance has rolled out a new AI product structure under the Binance Intelligence label, covering Binance AI for general users, Binance AI Pro for advanced users, and Agent OS for developers. At the launch event, the company said AI Pro is designed to turn a user’s natural-language trading idea into a workflow that can be edited, simulated, and only then pushed into live trading if the user approves it.
Three layers in the Binance Intelligence lineup
According to the official breakdown, Binance Intelligence is split into three parts.
- Binance AI: built for general users, free to use, and able to produce briefings on crypto, U.S. stocks, and macro markets. The content can also be adjusted based on a user’s habits, preferences, and holdings.
- Binance AI Pro: aimed at more advanced users and focused on converting natural-language prompts into automated workflows.
- Agent OS: built for developers, allowing external AI agents to connect with Binance data and trading capabilities.
During the event, Binance CEO Richard Teng and Vice President of Product Jeff Li said users only need to describe a trading idea in one sentence, after which AI Pro can organize it into a workflow, let the user revise it, test it, and decide whether to move it into live trading. Binance summarized that process as Prompt, Test, Trade.
The launch event also showed non-trading use cases, including automatically generated market reports and BTC monitoring with alerts at key price levels.
Test case one: using AI Pro to filter Binance Earn products
Foresight said it chose not to repeat basic checks such as market quotes and general data lookups, since earlier versions of Binance AI and the ChatGPT plugin already handled those tasks. Instead, it started with a more practical yield use case. Because the new AI Pro is connected to Binance Earn, the tester asked how 50 BTC could be allocated in Binance Earn to generate yield without taking principal-loss risk, while prioritizing products with higher current returns and lower risk.
According to the report, AI Pro directly pulled current Binance Earn product data and also read the user’s real account balance. Since the tester’s account only held a small amount of BTC, the tool proactively pointed out that there were not actually 50 BTC in the spot account and that BTC would need to be transferred in before any such plan could be executed.
On product selection, AI Pro recognized the request to avoid principal-loss risk and excluded products such as Dual Investment that could change the amount of BTC held. It prioritized BTC flexible and fixed-term products instead. The tool found a BTC flexible APR of about 0.26%, along with a 5-day promotional fixed-term product offering up to 20% APR. Foresight said those results were broadly consistent with what it later saw on the Earn page.
Test case two: building a BTCUSDT perpetual long strategy from a prompt
The second test focused on what Foresight described as AI Pro’s key feature: strategy creation. To see whether the system could handle more complex trading logic, the tester did not use a simple breakout-buy prompt. Instead, it combined several types of data commonly used in derivatives trading.
The prompt was: create a BTCUSDT perpetual long strategy where 1h and 4h are both above MA20; 15m reclaims MA20 and trading volume is more than 1.3 times the average of the past 20 candles; open interest rises more than 5% over the past hour; taker buy/sell ratio is above 1; and funding rate is below 0.01% at entry. Position size is 10%, leverage is 2x, stop-loss is 1.5%, take-profit is 3%, and the strategy should be generated without going live.
Foresight said AI Pro handled the logic better than expected. It split the structure into layers: 1h and 4h to determine the broader trend, 15m to identify the entry point, volume to confirm expansion, open interest and the taker buy/sell ratio to check whether capital flow was aligned, and funding to filter out overheated conditions.
The report also noted that execution details were handled with some care. For example, when the user specified a 10% position, AI Pro explained how it interpreted that value, including whether it referred to notional exposure or margin usage. It also surfaced additional settings that had not been fully specified, such as isolated margin, one-way position mode, and checking frequency, and asked the user to confirm them before proceeding.
Once the strategy was confirmed, AI Pro could send it into paper trading first. The tester did exactly that, placing the setup into simulated trading so the system could keep checking the rules against live market conditions. That was used both to see whether the strategy would trigger normally and to spot any flaws in the logic or parameter choices ahead of any live deployment. The article said the strategy is currently running in simulation, with any move to live trading left for later if performance remains stable and trigger behavior matches expectations.
The test focused on real usage rather than conversational ability
Foresight wrote that the main question was no longer whether an AI tool could chat. The more important issue was whether it could understand what the user wanted, call real-time data and account information, and then turn that into a strategy or an executable task. With that in mind, the test centered on two scenarios the writer said were genuinely relevant in daily use: yield allocation and trading strategy construction.
Foresight’s conclusion on where AI fits in trading
After the two rounds of testing, Foresight said the new AI Pro felt better than expected. The article argued that in AI trading, the more obvious strength lies in monitoring markets around the clock, checking data, waiting for conditions, and executing a strategy. In that view, those are the tasks that suit AI best.
The piece also pointed to Binance’s built-in advantage. It said the exchange has trading depth, a large user base, and substantial first-hand trading data. Many third-party AI tools still need to connect their own market data, account systems, and trading interfaces, while Binance can place AI directly inside its existing trading system.
The article’s final takeaway was straightforward: in the writer’s view, AI has started to enter the trading era in a more concrete way.
Disclaimer
The original article said markets carry risk and investing requires caution. It added that the piece does not constitute investment advice, and that users should consider whether any opinion, view, or conclusion in the article fits their own situation before making investment decisions at their own responsibility.
